JP3530474B2 - Wing vibration measurement method and wing vibration monitoring system using the same - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a blade vibration measuring method and a blade vibration monitoring system using the same, and more specifically, it uses a small number of sensors for the amplitude, phase and frequency of blade vibration generated in a turbine blade or the like. The present invention relates to a blade vibration measuring method for measuring and a blade vibration monitoring system using the same.

[0002]

2. Description of the Related Art Non-contact blade vibration measuring methods for measuring blade vibrations generated in turbine moving blades using a non-contact type sensor are generally roughly divided into non-contact type sensors for the entire circumference of the blade. Dozens of points are attached, a method for obtaining the amplitude, phase, and frequency of the vibration of each blade based on the detection signal of each blade by these sensors, and a non-contact type sensor is attached to the outer circumference of the blade at one or two points. There has been a method of obtaining only the amplitude of vibration of each blade based on the detection signal of each blade by a sensor.

[0003]

Therefore, conventionally, when not only the amplitude of blade vibration but also the phase and frequency are to be obtained, it is impossible with a small number of non-contact type sensors, and several dozen or more non-contact type sensors are used. It was necessary to install the sensor all around the wing. For this reason, a large amount of cost is required for measuring equipment such as a non-contact type sensor, and much labor and cost are required for sensor mounting work and maintenance.

Therefore, in view of the above problems, the present invention uses a blade vibration measuring method capable of obtaining the amplitude, phase, and frequency of blade vibration of a rotor such as a turbine rotor using a small number of sensors. An object is to provide a blade vibration monitoring system.

[0005]

[Means for Solving the Problems] First to solve the above problems
The blade vibration measuring method of the invention is a method for measuring the vibration of a blade attached to a rotating body, wherein the blade transit time is calculated based on the detection signals of at least three non-contact blade sensors arranged on the outer periphery of the blade. First step of measurement (step S in FIG. 3)
1), the second step of calculating the difference between the blade passing times (step S2 in FIG. 3), and the third step of removing the DC component (component generated between the blade sensors) from the blade passing time difference. 3) and a fourth step (FIG. 3) for obtaining the amplitude A, the phase C, and the order ratio B of the blade vibration by performing frequency analysis on the blade passage time difference from which the DC component has been removed by frequency analysis means. Step S4, Step S
5), and a fifth step (steps S6 and S7 in FIG. 3) of classifying blade vibration into an asynchronous component, a synchronous component, and a 0.5H component based on the order ratio B. (Step S8 in FIG. 3), a curve fit is applied to the equation of the phase C, and the order ratio n of the blade vibration and the phase φ
Sixth step for obtaining and (step S8-1 in FIG. 4)
And a seventh step (step S8-2 in FIG. 4) of obtaining the amplitude a of the blade vibration by using the order ratio n obtained in the sixth step in the equation of the amplitude A, and in the case of the synchronous component (Step S9 in FIG. 3), an eighth step (FIG. 5) in which the amplitude a and the phase φ of the blade vibration are obtained by performing a curve fit to the equation of the two-point method using the order ratio n as a parameter.
Step S9-1) and the ninth step (step of FIG. 5) of selecting the amplitude a and the phase φ by selecting the optimum condition from the order ratio n given as a parameter in the eighth step. S9-2) and 0.5H component (step S10 in FIG. 3), the amplitude a and the phase φ of the blade vibration are obtained by performing a curve fit to the equation of the two-point method using the order ratio n as a parameter. The amplitude a and the phase φ are selected by selecting the optimum condition from the tenth step (step S10-1 in FIG. 6) for obtaining and the order ratio n given as a parameter in this tenth step. An eleventh step (step S10-2 in FIG. 6),
It is characterized by having.

The blade vibration measuring method of the second invention is the first method.
In the blade vibration measuring method of the invention, the frequency analyzing means in the fourth step obtains the amplitude A, the phase C, and the order ratio B of the blade vibration by performing a Fourier transform on the blade passage time difference from which the DC component has been removed. With the twelfth step (step S4 in FIG. 3) and the blade passage time difference from which the DC component is removed, with the amplitude A, the phase C, and the order ratio B of the blade vibration obtained in the twelfth step as initial values, A thirteenth step (step S5 in FIG. 3) of performing curve fitting to obtain the amplitude A, the phase C, and the order ratio B of the blade vibration.

The blade vibration measuring method of the third invention is the second method.
In the blade vibration measuring method of the invention, in the case of an asynchronous component, after the seventh step of the first invention, a curve fitting condition Z of the order ratio is created by using the order ratio B obtained in the thirteenth step. 14th step (step S8-3 in FIG. 3) and a fifteenth step (FIG. 3 in FIG. 3) for obtaining the amplitude a and the phase φ of the blade vibration by performing curve fitting based on the curve fitting condition Z. Step S8
-4), and the 16th step of selecting the optimum condition of the curve fitting condition Z from the result of the curve fitting, and selecting the order ratio n of the blade vibration, the amplitude a, and the phase φ (step S8- in FIG. 3). 5) and are included.

Further, the blade vibration measuring method of the fourth invention is the blade vibration measuring method of the first, second or third invention, in which the predicted value of the static test by numerical analysis or tapping is used when curve fitting is performed. It is characterized by being reflected as a boundary condition.

The blade vibration measuring method of the fifth aspect of the invention is a method for measuring the vibration of a blade attached to a rotating body, wherein, instead of the first to third steps of the first aspect of the invention, the outer circumference of the blade is equalized. Seventeenth, which measures the blade transit time of each blade based on the detection signals of at least four non-contact blade sensors arranged in the pitch
(Step S21 in FIG. 14), the eighteenth step (step S22 in FIG. 14) for calculating the difference between blade passing times, and the nineteenth step for calculating the difference in blade passing time difference to remove the DC component. Step (step S2 in FIG. 14)
3), and the algorithm after the fourth step of the first invention is characterized by performing the algorithm of the first, second, third or fourth invention.

A blade vibration measuring method according to a sixth aspect of the present invention is a method for measuring the vibration of a blade attached to a rotating body, which is arranged on the outer periphery of the blade instead of the first and second steps of the first aspect of the invention. 20th step (step S31 of FIG. 17) of measuring the passage time of the blade based on the detection signals of the at least two non-contact blade sensors, and from this blade passage time, the reference point of rotation is always used. The 21st step of removing the DC component by taking the time difference (step S3 of FIG. 17)
2) and, and the algorithm after the third step of the first invention is characterized by implementing the algorithm of the first, second, third or fourth invention.

Further, the blade vibration measuring method of the seventh invention is a blade vibration measuring method when the blades attached to the rotating body vibrate in a coupled state, and at least three blades are arranged on the outer circumference of the blade. A first step (step S51 in FIG. 25) of measuring the passage time of the blade based on the detection signal of the non-contact blade sensor at the point and a second step (FIG. 25) of removing the DC component from the passage time of the blade. 25) and the blade passage time from which the DC component has been removed, frequency analysis is performed to search for a peak value, and then a third step (step S53 in FIG. 25) of confirming the phase difference for each peak. )
And a fourth step (step S54 in FIG. 25) of calculating the order ratio n based on this phase difference, and the order ratio,
Combined with the results of the frequency analysis, the Nodal diamond number N
Fifth step of calculating d (step S5 of FIG. 25)
5), a sixth step (step S56 in FIG. 25) of frequency analysis of the passage time of each blade, and a seventh step (step of FIG. 25) of developing the spectrum calculated by this frequency analysis in consideration of folding back. S57) and the eighth step of searching the amplitude and phase of the vibration of each blade having the highest peak in the order ratio n calculated in the fourth step based on the developed spectrum (step S58 in FIG. 25).
And the phase of each peak theoretically obtained is compared with the actual phase obtained in the eighth step to obtain a correct order ratio n, ninth step 8 (step S59 in FIG. 25).
Then, the order ratio n of each peak obtained in the ninth step is given as an initial value to the equation of the nodal diameter mode method to perform curve fitting, thereby obtaining the most accurate order ratio n.
0 step (step S60 of FIG. 25) and this first
The eleventh step (FIG. 25) of obtaining the amplitude a and the phase φ of the vibration of each blade based on the order ratio n obtained in the step 0
And step S61).

Further, the blade vibration measuring method of the eighth invention is a vibration measuring method of a blade attached to a rotating body in a coupled state, wherein the blade outer circumference is replaced by the first step of the seventh invention. Based on the detection signals of at least three non-contact blade sensors arranged at equal pitches, the twelfth step of measuring the blade passage time (step S71 in FIG. 34) and the difference between the blade passage times are calculated. And a thirteenth step (step S72 in FIG. 34), and the algorithm after the second step of the seventh invention is characterized by implementing the algorithm of the seventh invention.

The blade vibration monitoring system of the ninth invention is
The blade vibration data measured by the blade vibration measuring method of the first, second, third, fourth, fifth, sixth, seventh, or eighth invention is transmitted to a monitoring device to monitor the blade vibration. It is characterized in that it is configured in.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

<Embodiment 1> ... Optimization Method I FIG. 1 is a sensor layout view of a blade vibration measuring system according to Embodiment 1 (optimization method I) of the present invention. As shown in the figure, in the turbine rotor 10, a large number (for example, 60) of moving blades 12 are provided on the outer peripheral surface of a rotating shaft 11 at regular intervals in the circumferential direction. Numbers (for example, numbers 1 to 60) are set in order from 1 for each turbine blade 12. A key groove 13 is formed on the outer peripheral surface of the rotating shaft 11.

On the outer periphery of the turbine blade 12, three non-contact blade sensors 14-1, 14-2, 14-3 are arranged at predetermined intervals in the circumferential direction. These blade sensors 14-1, 14-2, 14-3 are proximity sensors, and the rotation of the turbine rotor 10 causes each turbine blade 12 to have its blade sensors 14-1, 14-2, 14-.
A blade detection signal is output at the time when the mounting position of No. 3 is reached (when the tip of the turbine blade faces the sensor). Further, a non-contact type key groove sensor 15 is arranged on the outer periphery of the turbine rotating shaft 11. The key groove sensor 15 is also a proximity sensor, and when the key groove 13 reaches the mounting position of the key groove sensor 15 by rotating the turbine rotor 10 (when the key groove faces the sensor), a key groove detection signal (rotation) is generated. Pulse) is output.

The blade sensors 14-1, 14-2, 14
-3 and the key groove sensor 15 are not limited as long as they can detect the turbine blade 12 and the key groove 13 in a non-contact manner, and an appropriate sensor such as an optical sensor can be used.

Sθ ₁ , Sθ ₂ and Sθ _{3 in} FIG. 1 are the mounting angles of the blade sensors 14-1, 14-2 and 14-3, respectively. That is, from the mounting position of the key groove sensor 15 (output position of the rotation pulse) to each of the blade sensors 14-1, 14-2,
It is an angle up to 14-3. B _N θ is the mounting angle of the Nth turbine blade 12. That is, it is the angle from the attachment position of the key groove 15 (output position of the rotation pulse) to the Nth turbine blade 12. For example, the first turbine blade 1
The mounting angle of 2 is B ₁ θ, the mounting angle of the second turbine blade 12 is B ₂ θ, and the mounting angle of the third turbine blade 12 is B
₃ θ.

FIG. 2 is a block diagram showing the configuration of a blade vibration measuring system according to the first embodiment (optimization method I) of the present invention. As shown in the figure, the blade sensors 14-1, 14-
The blade detection signals d _1N, d _2N and d _3N of ₂ , 14-3 are the clock 16
-1, 16-2, 16-3 are input respectively. The key groove detection signals (rotational pulses) e of the key groove sensor 15 are clocks 16-1, 16-2, 16-3 and clock 17 respectively.
Entered in and. In the watches 16-1, 16-2, 16-3, when the first key groove detection signal e is input from the key groove sensor 15, this is used as an opportunity to continuously measure the time thereafter. And the clocks 16-1, 16-2, 1
In 6-3, blade detection signal from the blade sensor 14-1, 14-2, 14-3 d _1N, d _2N, each time entering the d _3N, enter the blade detection signal d _1N, d _2N, the d _3N The time (blade passage time) T _1N , T _2N , and T _3N at the time of the output is output to the CPU 18.
That is, the blade passing times T _1N , T _2N , and T _3N are the blade sensors 14-1 and 14- once for each rotation of the turbine blade 12.
It is time to pass through 2, 14-3.

When it is not necessary to identify which turbine blade vibration, the key groove detection signal may not be used and a signal that triggers measurement may be simply given. Also, when identifying the turbine blade, the identification is not necessarily limited to the key groove detection signal, and other appropriate identifying means may be used. For example, one of the turbine blades may be processed and a reference pulse signal may be output when the blade sensor detects the processed portion.

_{N in} the blade detection signals d _1N , d _2N , d _3N and blade passage times T _1N , T _2N , T _3N is a blade number. Correspondence between the blade detection signals d _1N , d _2N , d _3N and each blade 12, that is, the blade passing time T _1N , T _2N , T _3N input to the CPU 18
The number of the turbine blade 12 depends on the sensor mounting angle S
Since it is uniquely determined by θ ₁ , Sθ ₂ , Sθ ₃ and the blade mounting angle B _N θ, this is stored in the CPU 18 in advance, and the CPU 18 makes the determination.

In the timepiece 17, the time from the input of the key groove detection signal e from the key groove sensor 15 to the input of the next key groove detection signal e, that is, the rotation cycle required for the turbine rotor 10 to make one rotation CPU 18 measures T _r sequentially
Output to. Note that the key groove detection signal e does not necessarily have to be used as the rotation reference signal, and an appropriate signal may be used as the rotation reference signal for _{obtaining the} rotation cycle T _r and the like.

As will be described in detail later, the CPU 1 uses the clock 1
Blade passing time T input from 6-1, 16-2, 16-3
_1N, T_2N, T_3NRotation cycle input from the clock or clock 17
T_rVibration generated in each turbine blade 12 based on
Amplitude a, phase φ, frequency f _bCalculate (rotational order ratio n)
To do.

The vibration amplitude a, the phase φ, and the frequency f _b of each blade 12 calculated by the CPU 18 are transmitted to the remote monitoring device 20 by the transmitter 19. In the remote monitoring device 20, the receiver 21 receives the amplitude a, the phase φ, and the frequency f _b of the vibration of each turbine blade 12 transmitted from the site, and CP
The signal is processed in U22 and displayed on the display 23. Therefore,
The operator can monitor the vibration state of each turbine blade 12 by viewing the amplitude a, the phase φ, and the frequency f _b of the vibration of each turbine blade 12 displayed on the display 23.

Here, the calculation processing of the vibration amplitude a, the phase φ, and the frequency f _b of the turbine blade 12 in the CPU 18 will be described in detail with reference to the flowcharts of FIGS.

First, the passage time T _1N of all the turbine blades 12,
T _2N and T _3N are measured (step S1 in FIG. 3). The specific measuring method is as described above. The blade passing times T _1N , T _2N , and T _3N measured here are expressed by the following equation (1),
Expressions (2) and (3) are obtained.

[0027]

[Equation 1]

These equations are known as one-point equations. Here, T _r is the rotor rotation period, that is,
This is the time required for the turbine rotor 10 to rotate once. The rotor rotation period T _r is measured based on the key groove detection signal e of the key groove sensor 15 as described above. Alternatively, the blade passage time for one rotation (if there are 60 turbine blades, the blade passage time at 60 points) may be integrated, or the blade passage time may be averaged over several rotations. r represents the rotor rotation, that is, the number of rotations of the turbine rotor 10. The rotor rotation r can be obtained by counting the number of inputs of the rotor rotation cycle T _{r in} the CPU 18. Alternatively, the rotor rotation r may be obtained by counting the key groove detection signal e of the key groove sensor 15. f
_b1 ~f _bx is the number of vanes vibration can be determined by calculating the product (n · Omega) and the rotation order ratio n and the rotor rotational speed Omega in CPU 18. Ω is the rotor rotation speed, and is the reciprocal of the rotor rotation cycle T _r (1
It can be obtained by calculating / T _r ). a
_{1 (f b1) ~a X (} f bx) is the amplitude of the blade vibration is a function given by the wing frequency f _b1 ~f _bx. φ ₁ (f _b1 ) ~ φ _X
(f _bx ) is the phase of the blade vibration, and is a function given by the blade frequencies f _{b1 to} f _bx .

U is a peripheral speed, which is obtained by calculating the product (Ω · L) of the rotor rotational speed Ω and the length L of the outer circumference (circumference drawn by the tip of the blade) of the turbine rotor 10 in the CPU 18. be able to. B _N θ and Sθ ₁ , Sθ ₂ ,
Sθ ₃ is determined by the turbine blade 12 and the blade sensor 14-
The mounting angles of 1, 14-2 and 14-3 are stored in advance in a memory or the like and used for the arithmetic processing in the CPU 18. n _{1 to} n _x are rotation order ratios, which are ratios (f _b / Ω) between the rotor rotation speed Ω and the blade vibration frequency f _b .

Then, in the above equations (1), (2) and (3), the first and second items are DC components, which are components that exist regardless of the presence or absence of blade vibration. When no blade vibration occurs, this DC component is the only component. And
The components after the third item in the equations (1), (2), and (3) are the components related to blade vibration.

FIG. 7 shows the passage times T ₁₁ and T of the first blade 12.
₂₁ and T ₃₁ are shown below. In FIG. 7, the vertical axis represents the angle θ (sensor mounting angle) and the horizontal axis represents the time T. Turbine rotor 1
Wing sensors 14-1, 14- once every 0 times
2,14-3 by being detected that flies wings 12, wings detection signal d _11, d _21, d ₃₁ is output. Then, the blade passing times T ₁₁ , T ₂₁ , T ₃₁ are sequentially measured based on these blade detection signals d ₁₁ , d ₂₁ , d ₃₁ . For example, the wing passage time T
Looking at ₁₁ is as shown in FIG. In FIG. 8, the vertical axis represents time T and the horizontal axis represents rotor rotation r. The number of measurement data points at this time is, for example, 32 points.

Next, the blade passing time T_1N, T_2N, T_3NDifference
Is calculated (step S2 in FIG. 3). That is, the wing sensor 1
Blade passing time difference (T) between 4-1 and blade sensor 14-2_1N
-T _2N), Between the wing sensor 14-2 and the wing sensor 14-3
Wing passage time difference (T_2N-T_3N), Wing sensor 14-1 and wing
Blade passage time difference (T_1N-T_3N)
Calculate each. Taking turbine blade 12 with blade number 1 as an example
For example, the wing passage time difference (T₁₁-T_{twenty one}), (T_{twenty one}−
T₃₁), (T₁₁-T₃₁) Is calculated. This wing passage time difference
(T₁₁-T_{twenty one}), (T_{twenty one}-T₃₁), (T₁₁-T₃₁) Is the expression
It can be expressed by the following equations (4), (5) and (6).
It

[0033]

[Equation 2]

These equations (4), (5) and (6) are known as two-point equations. (4),
In equations (5) and (6), one item is the DC component,
The following items are components related to blade vibration. By obtaining such a blade passing time difference, the above (1), (2),
It is possible to remove one item of the DC component (T _r · r) in the equation (3). At this time, for example, when (T ₁₁ −T ₂₁ ) is shown in FIG.
It is divided into three cases of (b) and (c).

FIG. 9A shows the case where no blade vibration occurs (hereinafter referred to as case I), and (T ₁₁ -T ₂₁ ) is the direct current component (T _r · r) / (of the above equation (4). It becomes a constant value corresponding to 2π). FIG. 9B shows the case where the blade vibration is generated and only the blade vibration component synchronized with the rotor rotation is present (hereinafter referred to as case II). In this case, the same value is measured every time for the actual vibration indicated by the dotted line. The DC level in this case is shown in FIG.
It is higher than the DC level before the blade vibration shown in (a). FIG. 9C shows the case where there is one blade vibration component that is not synchronized with the rotor rotation (hereinafter referred to as case III).
In this case, it is observed as a (low-frequency) sinusoidal blade vibration that changes more slowly than the actual blade vibration shown by the dotted line.

Then, (T ₁₁ −T ₂₁ ), (T ₂₁ −
From the values of (T ₃₁ ) and (T ₁₁ −T ₃₁ ), the DC component (T _r
r) / (2π) is removed (step S3 in FIG. 3). As a result, only the blade vibration component is obtained. There are the following methods (1), (2), and (3) as methods for removing the DC component. (1) Method of removing by theoretical value. (2) A method of removing the blade vibration based on measurement data at the timing when it does not occur. (3) A method of calculating a moving average value of actual measurement data and removing the moving average value.

In the method (1), the DC component (T_r・
r) / (2π) is calculated, and this value is calculated as (T₁₁−
T_{twenty one}), (T_{twenty one}-T₃₁), (T₁₁-T₃₁) Value
It is a method of pulling. In this calculation, the measured
Data rotation cycle T_rAnd the rotor rotation r. the above
The method of (2) is used when a certain blade is not vibrating.
(T₁₁-T_{twenty one}), (T_{twenty one}-T₃₁), (T₁₁-T₃₁) Total
Obtain the measured value, and generate rotor vibration Ω and blade vibration at this time.
Measurement by the ratio with the rotor speed Ω
Correct the value and correct the measured value
Measurement data (T₁₁-T _{twenty one}), (T_{twenty one}-T₃₁),
(T₁₁-T₃₁) Is the method of subtracting from the value.

The above method (3) uses the measurement data (T₁₁-T
_{twenty one}), (T_{twenty one}-T₃₁), (T₁₁-T ₃₁) Moving average
By calculating the value of the DC component, this moving average
Value (DC component value), measured data (T₁₁-T_{twenty one}), (T
_{twenty one}-T₃₁), (T₁₁-T₃₁) Value is subtracted from
It In this case, a simple moving average may be calculated and weighted
You may be hurt. That is, a known moving average method is appropriately used
be able to. This method using the moving average method
Accurate change of gradual DC component due to thermal deformation of blade
Can be captured by using the moving average method.
It is most desirable to remove the components.

Then, a DC component is formed by any of the above methods.
Three measurement data (T ₁₁-T_{twenty one}), (T_{twenty one}
-T₃₁), (T₁₁-T₃₁) Frequency analysis
By calculating the amplitude A, phase C, and rotational order ratio B of the blade vibration,
A peak search is performed (step S in FIG. 3).
4). In other words, how many (how many)
Whether the order ratio B is included, that is, the measured blade vibration is
Find out how many kinds of frequencies are combined. Lap
As a wave number analysis method, for example, FFT (Fast Fourier T
ransform) is used. However, (T₁₁-T_{twenty one}), (T
_{twenty one}-T₃₁), (T₁₁-T₃₁) Has three vibration waveforms
Therefore, FFT is performed for each and the average of all data
Or find the peak from the product.

FIGS. 10A, 10B, and 10C show examples of frequency analysis in cases I, II, and III, respectively.
In case I shown in FIG. 10A, no peak appears. In case II shown in FIG. 10B, the rotation order ratio B is 0.
A peak appears at the H position. In case III shown in FIG. 10C, a peak appears at the position of the rotational order ratio B other than 0H. However, the rotational order ratio B here is 0.5 H at maximum because it is observed as a waveform that causes aliasing (folding phenomenon). That is, since each blade sensor 14-1, 14-2, 14-3 measures each blade 12 only once while the turbine rotor 10 makes one rotation, from the sampling theorem, the maximum rotation order ratio is It becomes 0.5. For example, the actual rotation order ratio is 2.4H,
In the case of 3.0H and 3.5H, the rotation order ratios obtained here are 0.4H, 0H and 0.5H.

[0041] Subsequently, the three measurement data obtained by removing the DC component _{(T 11 -T 21), (} T 21 -T 31), (T 11 -T 31)
By performing a curve fit by the nonlinear least squares method, the accurate amplitude A, phase C, and rotation order ratio B of the blade vibration are obtained (step S5 in FIG. 3). That is, since the frequency analysis (FFT) has poor resolution, accurate fitting of the order of rotation B, amplitude A, and phase C is performed by performing curve fitting. Specifically, the measurement data (T ₁₁
Curve fitting is performed on the following equations (7), (8), and (9) regarding −T ₂₁ ), (T ₂₁ −T ₃₁ ), and (T ₁₁ −T ₃₁ ). However, the initial value at this time is the frequency analysis (F
The value obtained by FT) is used.

[0042]

[Equation 3]

It should be noted that Y is the number of peaks searched in the frequency analysis (FFT), that is, the number of rotation order ratios B. Further, the rotation order ratios B _{1 to BY} are 0.5H at maximum due to aliasing.

Next, according to the value of the rotation order ratio B of each peak, there are classified into three cases of a synchronous component, a 0.5H component and an asynchronous component. Specifically, it is determined whether the rotation order ratio B is 0.02H or less (step S6 in FIG. 3), and if it is 0.02H or less, it is determined to be a synchronous component (step S9 in FIG. 3). . When the rotation order ratio B is larger than 0.02H, it is further judged whether or not it is 0.48H or more (step S7 in FIG. 3), and when it is 0.48H or more, the 0.5H component Is determined (step S1 in FIG. 3).
0). Then, when the rotation order ratio B is larger than 0.02H and smaller than 0.48H, it is determined to be an asynchronous component.

In the case of the synchronous component (case II), FIG.
As shown in (b), since only the DC level is present, the phase C becomes all 0, and therefore the solution cannot be obtained by the same method as in the case of the asynchronous component. Also, 0.5
The H component is a special case of Case III, and in this case, due to the relationship of the sampling number (measurement data number), the accuracy is poor with the same method as the case of the asynchronous component. That is, 0.5H
In the case of the component, the measured data have the same value one by one as shown in FIG. 11, and therefore the same amplitude A and phase C cannot be obtained by the same method as in the case of the asynchronous component. Therefore, in the case of the asynchronous component and the synchronous component and 0.
The amplitude a, the phase φ, and the rotational order ratio n of the blade vibration are obtained by different methods for the case of the 5H component. The determination conditions for the synchronization component and the 0.5H component are not limited to 0.02H and 0.48H, and can be appropriately selected.

(Asynchronous component) In the case of an asynchronous component, first, the rotational order ratio n and the phase φ are obtained by curve fitting the equation of the phase C by the method of least squares (step S8- in FIG. 4). 1). That is, the above equations (7), (8), and (9) that are curve-fitted in step S5 of FIG. 3 and the above equations (4), (5), and (6) that are two-point equations Comparing the above, there is a relationship of the following expressions (10), (11), and (12) for the phase C.

[0047]

[Equation 4]

Therefore, for example, C₁₁, C_{twenty one}, C₃₁about
Looking at it, y = ax + b (however, a and b are constants, x is a parameter
Because of the linear form of the₁₁，
C_{twenty one}, C ₃₁For the equation of, also curve fit (least squares
(Approximation)₁And phase φ₁When
Ask for. However, C₁₁, C_{twenty one}, C₃₁Is a phase, so
These 3 points can be made by returning to the original when the value exceeds 2π.
It is necessary to correct it so that it becomes a straight line. That is,
C₁₁, C_{twenty one}, C₃₁Should be a straight line, but in Figure 11
As shown for example C₃₁Value of slope n₁, Intercept φ₁(f_b1)
If it is far from the straight line,₃₁2π
By adding, the correction is performed so as to approach the straight line.
Similarly, C₁₂, C_{twenty two}, C₃₂, ..., C_1Y, C_2Y，
C_3YThe curve fit (least squares approximation)
The rotation order ratio n₂~ N_YAnd phase φ₂~
φ_YAnd ask.

Subsequently, the rotation order ratio n ₁ obtained by the above curve fit (step S8-1 in FIG. 4) is applied to the expression of the amplitude A.
~ N _Y is used to find the amplitude a (step S8-
2). That is, the above equations (7), (8), and (9) that are curve-fitted in step S5 of FIG. 3 and the above equations (4), (5), and (6) that are two-point equations Are compared with each other, the following (13), (14), (1
There is a relation of formula 5).

[0050]

[Equation 5]

Therefore, these equations of the amplitude A are added to the above-mentioned steps.
Rotation order ratio n obtained in step S8-1 ₁~ N_YEach
Substitution and curve fitting by the least squares method
Or by the average (eg amplitude A₁₁, A_{twenty one}, A₃₁of
Amplitude a obtained from each equation₁By calculating the average of
Amplitude a₁~ A_YAsk for.

Even with the processing up to this point, the amplitudes a _{1 to} a _Y , the phases φ _{1 to} φ _Y , and the rotational order ratios n _{1 to} n _Y when the blade vibration is an asynchronous component can be obtained. Adds the following processing.

First, the curve fitting condition (order ratio condition)
Z is created (step S8-3 in FIG. 4). Specifically, since the rotation order ratio B obtained in step S5 of FIG. 3 is obtained with higher accuracy than the rotation order ratios n _{1 to} n _Y obtained in step S8-1 of FIG. Order ratio B
Is used to create a condition such as the following expression (16). Since the order ratio B is included in all of the above (7), (8), and (9), it is obtained accurately because it is averaged in the curve fit, while the phase C is individually calculated in each equation. Therefore, the accuracy of curve fitting becomes relatively poor. For this reason, the accuracy of the order ratio n obtained based on the expression of the phase C is relatively low.

[0054]

[Equation 6]

In the above equation, "truncation" is a decimal point or less.
Means rounding down, and "rounding up" is the decimal point
It means to round up the bottom. For example, n₁But
If 2.4H, round down (n₁) Is 2, cut up
Ge (n₁) Becomes 3. Also, "+ B" is the rotation order ratio
It means that the value below the decimal point of B is added, and "-
"B" is minus the value below the decimal point of the rotation order ratio B.
It means that. Therefore, for example, the rotation order ratio n₁But
2.4, Rotational order ratio B₁Rounded down if is 2.7
(N₁) + B₁Is 2.7, rounded down (n₁) -B₁Is
1.3, rounded up (n ₁) + B₁Is rounded up to 3.7
(N₁) -B₁Is 2.3.

And Z₁~ Z_YOrders combined by
Ratio conditions (eg Z₁= Rounded down (n ₁) + B₁, Z₂=
Round up (n₂) -B₂, ..., Z_Y= Rounded down (n
_Y) -B_Y2) for all
The following equations (17), (18), and (19), which are point method equations, are
By performing a curve fit by the nonlinear least squares method
The amplitude a of the blade vibration₁~ A_YAnd phase φ₁~ Φ_YAnd calculate
(Step S8-4 in FIG. 4). That is, (1
Rotational order ratio n in equations 7), (18) and (19)₁~ N _YTo
Rotational order ratio Z₁~ Z_YOrder the values of each combination of
Next, perform curve fitting by substituting. In addition, this curve
In step S8-1, steps S8-1 and S8-2 in FIG.
Amplitude a obtained in₁~ A_YAnd phase φ₁~ Φ_YAnd are initial values
And give.

[0057]

[Equation 7]

Incidentally, these (17), (18), (1
Equation (9) is obtained by removing the DC component (T _r · r) / (2π) in the above equations (4), (5) and (6), and these (17), (18) and (19) ) In (T ₁₁
As the values of −T ₂₁ ), (T ₂₁ −T ₃₁ ), and (T ₁₁ −T ₃₁ ), those obtained by removing the value of the DC component from the measured value are used.

Then, the optimum condition is selected from all combinations of the rotational order ratios Z _{1 to} Z _Y (step S8-5 in FIG. 4). That is, in step S8-4 of FIG. 4, by comparing the results of performing curve fitting under the condition that all combinations of the rotation order ratios Z _{1 to} Z _Y are compared,
A combination of rotation order ratios Z _{1 to} Z _Y having the smallest squared error is selected, and this is set as the rotation order ratios n _{1 to} n _Y.
And, the solution by selecting an amplitude a ₁ ~a _Y and the phase phi ₁ to [phi] _Y calculated in this condition.

(In the case of the synchronous component) In the case of the synchronous component,
First, the amplitude a and the phase φ of the blade vibration are obtained by curve-fitting the equation of the two-point method by the nonlinear least squares method using the rotational order ratio n as a parameter (step S9-1 in FIG. 5). That is, the following (2
0), (21), (22) in the rotational order ratio n, for example 1
Amplitude a and phase φ are obtained by sequentially substituting values up to -20H (values equal to or higher than the rotational order ratios that are expected to occur in numerical analysis or tapping tests) and performing curve fitting. Since the condition for determining the synchronization component in step S6 of FIG. 3 is 0.02H, n = 1 ±
By performing curve fitting with 0.02, 2 ± 0.02, ..., NH ± 0.02 (for example, N = 20) as the condition of the convergence range, the amplitude a and the phase at which the square error is minimized Find the value of φ.

[0061]

[Equation 8]

Incidentally, these (20), (21), (2
Equation (2) is obtained by removing the DC component (T _r · r) / (2π) in the above equations (4), (5), and (6), and these (20), (21), (22) ) In (T ₁₁
As the values of −T ₂₁ ), (T ₂₁ −T ₃₁ ), and (T ₁₁ −T ₃₁ ), those obtained by removing the value of the DC component from the measured value are used.

Then, the optimum condition is selected from the rotation order ratio n given as a parameter (step S9-2 in FIG. 5). That is, in step S9-1 of FIG. 5, by comparing the results of curve fitting under the condition of the rotational order ratio n, the value of the rotational order ratio n with the smallest squared error is selected, and under this condition. The calculated amplitude a and phase φ are selected as a solution.

(For 0.5H component) The calculation method for the 0.5H component is the same as for the synchronous component. Specifically, first, the amplitude a and the phase φ of the blade vibration are obtained by curve-fitting the equation of the two-point method by the nonlinear least squares method using the rotational order ratio n as a parameter (FIG. 6).
Step S10-1). That is, the rotational order ratio n of the equations (20), (21), and (22), which is the equation of the two-point method, has a value of, for example, 0.5 to 20.5H. A value equal to or greater than the expected rotation order ratio) is sequentially substituted to perform curve fitting to obtain the amplitude a
And the phase φ. However, since the determination condition of the 0.5H component in step S7 of FIG. 3 is 0.48H, n = 0.5 ± 0.02, 1.5 ± 0.0 here.
, ..., NH ± 0.02 (for example, N is 20.5) is used as the condition of the convergence range, and curve fitting is performed to obtain the values of the amplitude a and the phase φ that minimize the square error.

Then, the optimum condition is selected from the rotation order ratio n given as a parameter (step S10-2 in FIG. 6). That is, the value of the rotation order ratio n having the smallest squared error is selected by comparing the results of curve fitting performed under the condition of the rotation order ratio n in step S10-1 of FIG. The calculated amplitude a and phase φ are selected as a solution.

In the case of the asynchronous component, in the case of the synchronous component or in the case of the 0.5H component, f _b = nΩ may be calculated when further determining the frequency f _b from the rotational order ratio n.

In the above description, the case of obtaining the vibration amplitude a, the phase φ, and the frequency f _b of the turbine blade 12 having the blade number 1 has been described as an example, but the above processing is performed for the turbine blades 12 having other blade numbers. If it is also performed, the vibration amplitude a, the phase φ, and the frequency f _b of all the turbine blades 12 can be obtained.

As described above, according to the first embodiment (optimization method I), only three blade sensors 14-1, 14- are provided.
The amplitude a, the phase φ, and the frequency f _b of the vibration of the turbine blade 12 can be obtained only by using 2, 14-3. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced. Furthermore, since the measurement time difference between the blade sensors is used, the blade sensors 14-1, 1
Even if there is uniform shaking in 4-2 and 14-3,
Vibration measurement can be performed accurately.

In the blade vibration measuring system using this blade vibration measuring method, the blade sensors 14-1, 14-
It is also possible to monitor the vibration of the turbine blade 12 by determining the blade vibration amplitude a, phase φ, and frequency f _b based on the blade detection signals d _1N , d _2N , and d _3N of ₂ , 14-3.

The blade sensors used in the above blade vibration measuring method are not necessarily limited to three points and may be four or more points. Since the curve fitting is performed as can be seen from the above calculation algorithm, the accuracy increases as the amount of measurement data increases. In other words, the wing sensor is 4
Since the data of the 2-point method increases as the number of points increases to 5 (for example, in the case of 2 points, (T ₁₁ −T
_{21), (T 21 -T 31} ), (T 31 -T 41), (T 11 -
The calculation accuracy of T ₃₁ ), (T ₁₁ −T ₄₁ ), (T ₂₁ −T ₄₁ ), the amplitude a, the phase φ, and the frequency f _b is improved. In other words, from the relationship between desired accuracy and cost,
The number of blade sensors may be appropriately selected.

At present, it is possible to predict what the blade frequency will be by a numerical analysis or a static test by tapping, and this predicted value is in good agreement with the result of actual vibration measurement. ing. Therefore, if this predicted value is reflected and given as the boundary condition (condition of the convergence range) of the rotational order ratio at the time of curve fitting in the above blade vibration measuring method, for example, at the time of curve fitting at step S8-4 in FIG. It is possible to improve accuracy.

In the above, the blade mounting angle B _N θ is also included in the solution. However, since this B _N θ is common to all equations, it is not necessary to solve it including B _N θ, and later You may correct | amend and you may obtain | require a phase. That is, after obtaining the phase φ by solving without including B _N θ, this phase φ is φ ′ = φ
Correction is performed by calculating + n · B _N θ to obtain the actual phase φ ′. It is conceivable that the correction of B _N θ afterwards in this way will speed up the processing and may obtain a more correct value, since the mathematical formula at the time of solution will be simpler.

<Embodiment 2> ... Optimization method II In the above-described Embodiment 1 (optimization method I), when the difference in blade passing time between blade sensors is obtained (see step S2 in FIG. 3). , DC component (T _r · r) / as component between blade sensors
Since (2π) remains (see equations (4), (5), and (6)), this component was removed by the moving average method or the like (see step S3 in FIG. 3). On the other hand, if the blade sensors have four points at equal pitches (intervals), the blade sensor components (T _r · r) / (2π) have the same value, so the blade passing time difference is simply calculated, and It is possible to obtain three equations in which the inter-blade sensor component (T _r · r) / (2π) is removed only by calculating the difference in blade transit time difference. That is, it is possible to obtain three measurement data from which the DC component has been removed. The optimization method II according to the second embodiment focuses on this point. The optimization method II will be specifically described below.

FIG. 12 shows the second embodiment (optimization method) of the present invention.
II) sensor layout of the blade vibration measurement system, FIG.
FIG. 6 is a block diagram showing a configuration of a blade vibration measurement system according to a second embodiment (optimization method II) of the present invention. 12 and 13, the same parts as those in FIGS. 1 and 2 are designated by the same reference numerals, and the duplicated description will be omitted.

As shown in FIGS. 12 and 13, in the second embodiment (optimization method II), four blade sensors 14-1, 1 are used.
4-2, 14-3, 14-4 are used. These four blade sensors 14-1, 14-2, 14-3, 14-4 are arranged at equal pitches (intervals). That is, the wing sensor 14-
The mounting angles of ₁ , 14-2, 14-3, and 14-4 (angles from the rotation pulse output position to each blade sensor) are Sθ ₁ , S
Letting θ ₂ , Sθ ₃ and Sθ ₄ be (Sθ ₄ −Sθ ₃ ) =
(Sθ ₃ −Sθ ₂ ) = (Sθ ₂ −Sθ ₁ ) = δ.

The fourth blade sensor 14-1 is also a non-contact sensor such as a proximity sensor like the other blade sensors, and the turbine blade 12 rotates in front of the blade sensor 14-4 by the rotation of the turbine rotor 10. The blade detection signal d _4N is output at the time when it passes (when the tip of the turbine blade faces the sensor). This wing detection signal _d4N is input to the clock 16-4.
The keyway detection signal (rotation pulse) e of the keyway sensor 15 is also input to the timepiece 16-4. In the timepiece 16-4, when the first keyway detection signal e is input from the keyway sensor 15 as in the case of other timepieces, this is used as a trigger to continuously measure the time thereafter. And with the clock 16-4,
Every time the blade detection signal d _4N is input from the blade sensor 14-4, the time (blade passage time) T _4N at the time when the blade detection signal d _4N is input is output to the CPU 18.

In the CPU 18, the clocks 16-1, 16-2,
Blade passing time T input from 16-3 and 16-4_1N, T
_2N, T_3NAnd T_4NRotor circumference input from the clock or clock 17
Period T _rThe vibration generated in each turbine blade 12 based on
Motion amplitude a, phase φ, frequency f_bCalculate (rotational order ratio n)
Put out. Here, based on the flowchart of FIG.
The vibration of the turbine blade 12 in the CPU 18
Width a, phase φ, frequency f _bWill be described in detail
It

First, the passage time T _1N of all the turbine blades 12,
T _2N , T _3N and T _4N are measured (step S2 in FIG. 14).
1). The specific measurement method is as described above,
The same as Embodiment 1 (Optimization method I) except that the number of measurement points is one more than that of Embodiment 1 (Optimization method I). The blade passing time T _4N measured here is also the same as in the above equations (1), (2), and (3).
It becomes like the formula 3).

[0079]

[Equation 9]

Subsequently, the blade passing times T _1N , T _2N , T _3N , T
The difference of _4N is calculated (step S22 in FIG. 14). That is,
Taking the turbine blade 12 having the blade number 1 as an example, there are five types of blade passage time differences (T ₁₁ −T ₂₁ ), (T ₂₁ −T ₃₁ ), and (T ₃₁
-T _41), and calculates the _{(T 11 -T 31), (} T 21 -T 41). Blade passing time difference _{(T 11 -T 21), (} T 21 -T 31),
When (T ₁₁ −T ₃₁ ) is represented by an equation, the above (4), (5),
The equation becomes a two-point method like the equation (6), and the blade passing time difference (T
_31- T ₄₁ ) and (T ₂₁ -T ₄₁ ) can also be expressed by the two-point method equations similar to the equations (4), (5), and (6), although a specific description is omitted. Incidentally, the difference S.theta ₁ sensor mounting angle at which the formula _{-Sθ 2, Sθ 2 -Sθ 3,} Sθ 3 -
Sθ ₄ has the same value (δ), and Sθ ₁ −Sθ ₃ and Sθ ₄
θ ₂ −Sθ ₄ also has the same value (2δ).

Then, the DC component (T_r・ R) / (2π)
In order to eliminate the
The difference between the teachers is calculated (step S23 in FIG. 14). That is,
(T ₁₁-T_{twenty one})-(T_{twenty one}-T₃₁), (T_{twenty one}-T₃₁) −
(T₃₁-T₄₁), (T₁₁-T₃₁)-(T_{twenty one}-T₄₁)
Calculate each. When these are expressed by equations, the following (24),
Expressions (25) and (26) are obtained.

[0082]

[Equation 10]

Thus, three measurements with the DC component removed
Data (T₁₁+ T₃₁), (T_{twenty one}+ T ₄₁), ((T₁₁-T
₃₁)-(T_{twenty one}-T₄₁)) And its formulas (24), (2
5) and (26) are obtained. Removal of these three DC components
Measurement data after leaving (T₁₁+ T₃₁), (T_{twenty one}+ T₄₁),
((T₁₁-T₃₁)-(T_{twenty one}-T₄₁)) Is a form of the above implementation
In step S3 of FIG. 3 in the state 1 (optimization method I)
Measurement data after removal of DC component (T₁₁-T_{twenty one}),
(T_{twenty one}-T₃₁), (T₁₁-T₃₁) Is equivalent to
It

Therefore, the measurement data (T ₁₁ + T ₃₁ ), (T ₂₁ + T ₄₁ ), ((T ₁₁ −
T ₃₁ ) − (T ₂₁ −T ₄₁ )) and its expressions (24) and (2
5) and (26), and thereafter using the same algorithm as in the first embodiment (optimization method I), that is, FIG.
By executing the algorithm after step S4 of FIG. 4 and the algorithms shown in FIGS. 4, 5, and 6, the vibration amplitude a, the phase φ, and the frequency f _b of the turbine blade 12 with the blade number 1 can be obtained (FIG. 14). Step S24). Of course, for the turbine blades 12 having other blade numbers, the same processing as the turbine blade 12 having the blade number 1 is performed to obtain the vibration amplitude a, the phase φ, and the frequency f _b of all the turbine blades 12. You can

As described above, the second embodiment (optimization method)
According to II), only four blade sensors 54-1, 54-
2, 54-3, 54-4 are used, the turbine blade 1
The vibration amplitude a, the phase φ, and the frequency f _b of 2 can be obtained. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced.

Further, in the blade vibration measuring system using this blade vibration measuring method, the blade sensors 54-1, 54-
2, 54-3, 54-4 blade detection signals d _1N , d _2N ,
It is also possible to monitor the vibration of the turbine blade 12 by _obtaining the blade vibration amplitude a, phase φ, and frequency f _b based on d _3N and d _4N .

<Third Embodiment> ... Optimization Method III In the above first embodiment (optimization method I), the DC component (T _r · T) is obtained by obtaining the difference in blade passage time between blade sensors.
r) has been removed (see equations (1) to (6), step S2 in FIG. 3). On the other hand, if the rotation pulse (key groove detection signal) is taken in and the time difference from the time when the rotation pulse is detected is always taken, the DC component (T _r · r) can be obtained without obtaining the blade passage time difference between the blade sensors. Can be removed. What paid attention to this is the optimization method II of the third embodiment.
It is I. The optimization method II will be specifically described below.

FIG. 15 shows the third embodiment (optimization method) of the present invention.
III) Sensor layout of blade vibration measurement system, Fig. 1
6 is a block diagram showing the configuration of a blade vibration measurement system according to a third embodiment (optimization method III) of the present invention. 12
In FIG. 13 and FIG. 13, the same parts as those in FIGS.

As shown in FIGS. 15 and 16, the third embodiment (optimization method III) has the same sensor arrangement and system configuration as those of the first embodiment (optimization method I). However, as will be described later in detail, the third embodiment (optimization method III
), The DC component (T
_r.r ) is removed, and the blade sensor 14-3 surrounded by the alternate long and short dash line in FIG. 15 and FIG. 16 can be deleted and only two blade sensors 14-1 and 14-1 can be used. This is different from the first embodiment (optimization method I).

Based on the flowchart of FIG. 17 and the like, C
The calculation processing of the vibration amplitude a, the phase φ, and the frequency f _b of each turbine blade 12 in the PU 18 will be described in detail.

First, the passage time T _1N of all the turbine blades 12,
T _2N and T _3N are measured (step S31 in FIG. 17). The specific measuring method is as described in the first embodiment (optimizing method I), and the blade passing time T
The equations of _1N , T _2N and T _3N are the equations (1), (2) and (3). Then, the rotation pulse (key groove detection signal) e is taken in, and the rotation pulse is always detected (that is, the rotation reference point).
The DC component (T _r · r) is removed from the measured blade passing times T _1N , T _2N and T _3N by taking the time difference from (step S32 in FIG. 17).

Specifically, every time the rotation pulse e is output from the key groove sensor 15, that is, every time the rotor rotation period T _r is input from the timepiece 17, the calculation of (T _r · r) is performed. The rotor rotation r is obtained by counting the number of times the rotor rotation cycle T _r is input as described above.

Taking the turbine blade 12 with the blade number 1 as an example, from the blade passing times T ₁₁ , T ₂₁ , T ₃₁ input from the clocks 16-1, 16-2, 16-3, this (T _r
By subtracting the value of r), the DC component (T _r · r)
To remove. The blade passing times T ₁₁ , T ₂₁ , and T ₃₁ after removing the DC component (T _r · r) are represented by the following (27) and (2).
Expressions 8) and (29) are obtained. These equations (27),
(28) and (29) correspond to the rotor rotation r set to 0 in the above formulas (1), (2) and (3). In addition,
As a method of removing the DC component (T _r · r), the method shown in FIG.
In the watches 16-1, 16-2, and 16-3 shown in FIG. 1, the rotation pulse e is always input from the key groove sensor 15, and the rotation pulse e is input for each rotation, and then the blade sensor 14-
The time from ₁ , 14-2, 14-3 until the blade detection signals d _1N , d _2N are input may be measured.

[0094]

[Equation 11]

In this case, in the equations (4), (5), and (6) of the first embodiment (optimization method I), (T _r · r) / (2π) remains as the DC component ( 9 (a)), (27), (28), and (29), as a DC component, ( _{Bn [} theta] -S [theta] ₁ ) / (2 [pi]), ( _{Bn [} theta] -S).
θ ₂ ) / (2π) and (B _n θ-Sθ ₃ ) / (2π) remain, and this point is the same as in the first embodiment (optimization method I) and the third embodiment (optimization method III). Different. FIG. 18 shows an example of the case of T _1n .

Therefore, although there are differences in the DC component, the measurement data T ₁₁ , T ₂₁ , T after the removal of these three DC components are performed.
₃₁ and its equations (27), (28), (29),
After that, if the same algorithm as in the first embodiment (optimization method I) is used, that is, after the processing of step S3 in FIG. 3 (DC component removal processing by moving average etc.) and in FIG. 4, FIG. 5, and FIG. By executing the algorithm shown, the vibration amplitude a, the phase φ, and the frequency f _b of the turbine blade 12 having the blade number 1 can be obtained (step S33 in FIG. 17). Of course, for turbine blades 12 having other blade numbers, blade number 1
By performing the same processing as the turbine blade 12 of No. 3, the vibration amplitude a, the phase φ, and the frequency f _b of all the turbine blades 12 can be obtained.

By the way, although the case of using the three-blade sensors has been described above, the third embodiment (optimization method) will be described.
III) can be implemented with a minimum of two wing sensors. That is, in the first embodiment (optimization method I), the blade passage time difference between the blade sensors (two-point method formula)
Since the direct current component (T _r · r) is removed by obtaining, at least three blade sensors are required.
In the third embodiment (optimization method III), the DC component (T
It is possible to use two blade sensors to eliminate _r / r). However, it is not always necessary to use the rotation pulse (key groove detection signal) as the rotation reference signal, and in addition, the time difference may be obtained by using an appropriate signal as the rotation reference signal.

The reason why at least two blade sensors are required (the reason why one blade sensor is not enough) is that the equation for phase C ((10), (1
This is because at least two equations are required in order to obtain both the rotational order ratio n and the phase φ by curve fitting by the least-squares method with respect to 1) and (12). This is also because when the curve is finally fitted, the curve is fitted to the sine wave. Here, in the case of the synchronous component, no matter how many rotation data are taken, it becomes a direct current component, so if there is only one blade sensor, the direct current component is observed, and from this value the amplitude and phase of the original sine wave, It is not possible to determine the frequency and at least two wing sensors are needed. 19 (a) when there is only one wing sensor
Only the information shown in (1) is obtained, and in some cases it becomes 0. On the other hand, if there are two or more wing sensors,
As shown in (b), the amplitude, phase, and frequency of the turbine blade can be estimated from the difference between the synchronous components.

By the way, when the blade sensor has two points, when the equation of phase C is curve-fitted to obtain the rotational order ratio n and the phase φ in step S8-1 of FIG. Therefore, the value of the phase C cannot be corrected as in the case of the first embodiment (optimization method I), that is, when the phase C has three points. Therefore, in reality, the measurement accuracy may deteriorate.

However, this can be dealt with by making the distances between the two blade sensors sufficiently close to each other. In other words, if the two wing sensors are brought close enough, the phase C values at the two points should also be very close values, so a phase difference of 2π or more will occur unless the vibration frequency is considerably high. Therefore, no correction is required.
In addition, the accuracy of measurement can be improved by reflecting the predicted value of the blade vibration frequency by the numerical analysis or the static test by tapping on the curve fit.

As described above, the third embodiment (optimization method)
According to III), at least two blade sensors 54-1, 54
The vibration amplitude a, the phase φ, and the frequency f _b of the turbine blade 12 can be obtained only by using −2. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced.

Further, in the blade vibration measuring system using this blade vibration measuring method, the blade sensors 54-1 and 54-2 are always operated.
Of the turbine blade 12 by obtaining the amplitude a, the phase φ, and the frequency f _b of the blade vibration based on the blade detection signals d _1N and d _2N of
Vibration monitoring can also be performed.

<Embodiment 4> ... Node-diameter mode method I In the optimization methods I, II, and III of the above-described Embodiments 1, 2, and 3, each turbine blade is attached to a rotating shaft independently of each other. However, the nodal diameter mode method I of the fourth embodiment and the nodal diameter mode method II of the next fifth embodiment are formed by connecting all turbine blades attached to the rotating shaft over the entire circumference. This method can be applied when vibrating in the spelling mode (vibrating in a coupled state).

FIG. 20 is a sensor layout diagram of the blade vibration measuring system according to the fourth embodiment (node diameter mode method I) of the present invention,
FIG. 21 is an explanatory diagram showing a structural example of a turbine blade.

As shown in FIG. 20, in the fourth embodiment (nodal diameter mode method I), two blade sensors 54-1 and 54- are used.
2 is used. Then, in the turbine rotor 50 to be measured here, a large number (for example, 60) of moving blades 52 attached to the rotating shaft 51 are welded to each other, or as shown in FIG. All turbine blades 52 are in a coupled state by, for example, having a structure in which the tip portions (head portions) of the above are intertwined with each other. In the turbine rotor 50 having such a structure, the turbine blades 52 vibrate in the all-peripheral spelling mode. That is, since all the turbine blades 52 are coupled, at the resonance point, all the turbine blades 52 have a mutual phase difference and simultaneously vibrate integrally.

Regarding the vibration in the all-round spelling mode,
A state at a certain moment can be expressed as shown in FIG. 23 or FIG. 23 and 24, +,
The − represents the swing to one side and the swing to the other side in the rotation axis direction, respectively.

The turbine blades 52 are sequentially numbered from 1 (for example, numbers 1 to 60).
A key groove 53 is formed on the outer peripheral surface of the rotating shaft 51.
Two non-contact blade sensors 5 are provided on the outer circumference of the turbine blade 52.
4-1 and 54-2 are arranged at predetermined intervals in the circumferential direction. These blade sensors 54-1 and 54-2 are proximity sensors, and the rotation of the turbine rotor 50 causes the respective turbine blades 52 to move to the respective blade sensors 54-1 and 54-2.
The blade detection signal is output at the time when the mounting position of 2 is reached (when the tip of the turbine blade faces the sensor). Further, a non-contact type key groove sensor 55 is arranged on the outer periphery of the turbine rotating shaft 51. This key groove sensor 55 is also a proximity sensor, and the key groove 53 formed on the rotary shaft 51 is rotated by the rotation of the turbine rotor 50 so that the key groove sensor 5
A key groove detection signal (rotation pulse) is output when the mounting position of No. 5 is reached (when the key groove faces the sensor).

The blade sensors 54-1 and 54-2 and the key groove sensor 55 may be anything that can detect the turbine blade 52 and the key groove 53 in a non-contact manner, such as an optical sensor.
Appropriate sensors can be used.

In FIG. 20, Sθ ₁ and Sθ ₂ are the mounting angles of the blade sensors 54-1 and 54-2, respectively. That is,
Mounting position of key groove sensor 55 (output position of rotation pulse)
To the blade sensors 54-1 and 54-2.
B _N θ is the mounting angle of the Nth turbine blade 52. That is, it is the angle from the attachment position of the key groove 55 (output position of the rotation pulse) to the Nth turbine blade 52. For example, the mounting angle of the _first turbine blade 52 is B ₁ θ, the mounting angle of the second turbine blade 52 is B ₂ θ, and the mounting angle of the third turbine blade 52 is B ₃ θ.

FIG. 22 is a block diagram showing the configuration of a blade vibration measuring system according to the fourth embodiment (node diameter mode method I) of the present invention. As shown in the figure, the blade sensors 54-1,
The blade detection signals d _{1N and} d _2N of 54-2 are clocks 56-1 and 56.
-2 are input respectively. The key groove detection signal (rotation pulse) e of the key groove sensor 55 is the clock 56-
1, 16-2 and the clock 57.

In the clocks 56-1 and 56-2, when the first key groove detection signal e is input from the key groove sensor 55, this is used as a trigger, and the subsequent blade sensors 54-1 and 54-4.
-2 based on the blade detection signals d _1N and d _2N
Time T during which 2 passes through the blade sensors 54-1 and 54-2 sequentially
₁ (N), T ₂ (N) are measured, and these blade passing times T
₁ (N) and T ₁ (N) are output to the CPU 58. That is, the blade passing times T ₁ (N) and T ₂ (N) are functions of the blade number N,
If the total number of turbine blades 52 is 60, for example, 60 points of measurement data can be obtained per rotation.

Blade detection signals d _1N , d _2N and blade transit time T
N in ₁ (N) and T ₂ (N) is a blade number. Correspondence between the blade detection signals d _1N and d _2N and each turbine blade 12, that is, the blade passing times T _1N , T _2N and T _{3N which} are input to the CPU 18 at what order.
No. of the wing 52 depends on the sensor mounting angles Sθ ₁ , S
Since it is uniquely determined by θ ₂ , Sθ ₃ and the blade mounting angle B _N θ, this is stored in the CPU 58 in advance, and the CPU 58 makes the determination.

In the timepiece 57, the time from the input of the key groove detection signal e from the key groove sensor 55 to the input of the next key groove detection signal e, that is, the rotation cycle required for the turbine rotor 50 to make one rotation The CPU 58 measures T _r sequentially
Output to.

As will be described in detail later, the CPU 58 uses the clock 5
Blade passing time T ₁ (N), T input from 6-1 and 56-2
₂ (N) or the rotor rotation period T _r input from the clock 57, the amplitude a of the vibration generated in each turbine blade 52, the phase φ, and the frequency f _b (rotation order ratio n) are calculated.

Each turbine blade 52 calculated by the CPU 58
The vibration amplitude a, the phase φ, and the frequency f _b are transmitted by the transmitter 59 to the remote monitoring device 60. Remote monitoring device 60
Then, the vibration amplitude a, the phase φ, and the frequency f _b of each turbine blade 52 transmitted from the site are received by the receiver 61,
The signal is processed by the CPU 62 and displayed on the display 63. Therefore, the worker is required to display each turbine blade 52 displayed on the display 63.
By looking at the vibration amplitude a, the phase φ, and the frequency f _b of
The vibration state of each turbine blade 52 can be monitored.

Now, the calculation processing of the vibration amplitude a, the phase φ, and the frequency f _b of the turbine blade 52 in the CPU 58 will be described in detail with reference to the flowchart of FIG.

First, the passage time T of all turbine blades 52
₁ (N) and T ₂ (N) are measured (step S5 in FIG. 25).
1). The specific measuring method is as described above. In the case of all-round spelling mode vibration, since the phase between all turbine blades is given by a sine wave, the blade passage time T measured here
₁ (N) and T ₂ (N) can be expressed by the following equations (30) and (3
It becomes like the formula 1). In addition, the blade passing times T ₁ (N), T
₂ (N) is measured data T _1N , T for each turbine blade 52.
If viewed as _2N , it is expressed as the following equations (32) and (33) (equations of the one-point method) (similar to the above equations (1) and (2)).

[0118]

[Equation 12]

[0119]

[Equation 13]

The above equations (30) and (31) are node diameter modes.
This is known as the law formula. Where T
₁(N), T₂(N) is the blade sensor 54-1, 5 as described above.
4-2 is the blade transit time measured by
It becomes a function. T_rIs the rotor rotation period, that is, the turbine
This is the time required for the motor 50 to make one rotation. This low
Rotation cycle T_rIs the key of the keyway sensor 55 as described above
The measurement is performed based on the groove detection signal e. r is the rotor rotation, immediately
The number of revolutions of the turbine rotor 50 is shown.
This rotor rotation r is the rotor rotation cycle in the CPU 58.
T_rCan be obtained by counting the number of times
it can. Alternatively, the keyway detection signal e of the keyway sensor 55
May be counted to obtain the rotor rotation r. f_b1~ F
_bxIs the blade frequency, and the rotation order ratio n in the CPU 58
By calculating the product (n · Ω) of
Can be requested. Ω is the rotor speed, CP
Rotor rotation period T at U58_rReciprocal of (1 / T_r)
It can be obtained by calculating. a₁(f_b1)
~ A_X(f_bx) Is the amplitude of the blade vibration, and the blade frequency f_b1~ F
_bxIs a function given by. φ₁(f_b1) ~ Φ_X(f_bx) Is
The blade vibration phase (mode phase), and the blade frequency f _b1~
f_bxIs a function given by. N is the wing number as described above
Is. N_maxIs the number of all turbine blades (eg 60)
Is.

U is the peripheral speed, which is obtained by calculating the product (Ω · L) of the rotor rotational speed Ω and the length L of the outer circumference of the turbine rotor 10 (circle drawn by the tip of the blade) in the CPU 18. be able to. B _N θ and Sθ ₁ , Sθ ₂ ,
Sθ ₃ is determined by the turbine blade 12 and the blade sensor 14-
The mounting angles of 1, 14-2 and 14-3 are stored in advance in a memory or the like and used for the arithmetic processing in the CPU 18. The blade passing times T ₁ (N) and T ₂ (N) are B
_N θ is not included. This blade passing time T ₁ (N), T
₂ (N) is not the result for each blade, but the result for all blades. Since it is assumed that the turbine blade swings in the nodal diameter mode, there is no B _N θ in the equations for the blade transit times T ₁ (N) and T ₂ (N). In addition, B _N θ is B _N θ = (N / N _max ).
This corresponds to the expression Nd ₁ + φ ₁ . Nd _{1 to} Nd _x
Is the Nd-order nodal diamond number (reverse wave mode when positive, traveling wave mode when negative). That is, the nodal diamond number Nd is a value representing how many sine waves are present in the all-round spelling mode vibration during one rotation of the turbine rotor 50. For example, Nd is 3 in the case of FIG. It becomes a nodal diamond. n _{1 to} n _x are rotational order ratios, which are the ratio of the rotor rotational speed Ω to the blade frequency f _b (f _b / Ω)
Is.

Further, in order to simplify the equations (30) and (31), N = N + in the equations (30) and (31).
Assuming rN _max , the blade transit times T ₁ (N) and T ₂ (N) are given by the following equations (34) and (35). This means that, for example, when the total number of turbine blades is 60, the blade number is originally 60, followed by 1, 2, ..., But 61, 62, ...・ It means that

[0123]

[Equation 14]

Then, the DC component is removed from the values of the blade passing times T ₁ (N) and T ₂ (N) (step S5 in FIG. 25).
2). That is, the DC component of the first item of the equations (34) and (35) is removed. The method of removing the DC component will be described in detail below. That is, the measurement result is shown in FIG.
Therefore, the trend component should be removed. To remove the trend component, follow (1) ~
There are three methods of (3).

(1) A first-order approximation curve (trend component) of the blade number and the measurement time is calculated for each rotation or more, and the difference from the result is calculated. (2) The result of the first-order approximation is interpolated using the rotation pulses of two or more points, and the difference from this result is obtained. That is, 0 on the horizontal axis
The trend component is obtained by taking the rotation angle of 360 ° and finding the slope when the vertical axis is the time, and the difference from this result is taken. The specific procedure is as follows (a)-
It is (b). (A) First-order approximation is performed with the time between rotation pulses set to 360 °. (B) The relationship between the rotation pulse and the blade position is determined theoretically or experimentally (using the moving average value from the previous rotation). For example, when the number of blades is 100, 0.1 deg, 3.7 d
eg, 7.3 deg, 10.8 deg, and so on. (C) The trend component is calculated by substituting the blade position with the inclination obtained from the rotation pulse. (D) Take the difference from this result. (3) A device such as a rotary encoder that can mechanically interpolate the rotation pulse is attached, and polynomial approximation of the pulse train output from this device is performed. Then, the trend component is calculated by substituting the above-mentioned blade position relationship into this result, and the difference from this result is obtained.

Blade passage time T after removal of DC component
₁ (N) and T ₂ (N) are expressed by the following equations (36) and (37). That is, the blade passing times T ₁ (N) and T ₂ (N) at this time
The value of is only the vibration component.

[0127]

[Equation 15]

Next, the blade passing time T which is only the vibration component
Frequency analysis (FFT) is performed on ₁ (N) and T ₂ (N) to search for a peak value (step S53 in FIG. 25).
At this time, as shown in FIG. 28, a peak appears in (n + Nd) harmonics. At this point, the amplitude a ₁ (total average amplitude) of the blade vibration and the phase φ ₁ (mode phase) are obtained. After that, the phase difference is confirmed for each peak (step S53 in FIG. 25). That is, the blade passing time T
The difference between the phases φ obtained by frequency analysis of ₁ (N) and T ₂ (N) is obtained. At this time, of course, since the phase difference is confirmed, T
Accurately match the blade numbers N of ₁ (N) and T ₂ (N). As is clear from the comparison between the equations (36) and (37),
The phase difference is expressed by the following equation (38).

[0129]

[Equation 16]

Therefore, from the equation (39) obtained by modifying the equation (38), the rotation order ratio n is calculated based on the phase difference.
Is calculated (step S54 in FIG. 25). However, since the phase difference is considered, the phase difference + 2π, the phase difference + 4π, ...
Phase difference −2π, phase difference −4π, ...
By substituting these phase differences into the equation (39), a plurality of types of rotational order ratios n are obtained.

Then, the order ratio n and the above step S
Nodal diamond number Nd together with the frequency analysis result of 53
Is calculated (step S55 in FIG. 25). That is, since a peak appears in (n + Nd) harmonics in the frequency analysis, the value of Nd is obtained by subtracting the value of the order ratio n obtained in step S54 from the order ratio (n + Nd) that was peak-searched by the frequency analysis. .

Here, frequency analysis (FFT) is performed on the blade passage time T of each turbine blade 52 represented by the one-point method formula (step S56 in FIG. 25). That is, for example, taking the turbine blade 52 of blade number 1 as an example, first, from the blade passing times T ₁₁ and T ₂₁ of the turbine blade 52 of blade number 1 measured in step S51, the above (32), ( The DC component (T _r · r) of the first item of the equation 33) is removed. The method of removing this DC component is the same as that in step S32 of FIG. The equations at this time are the following equations (40) and (41). Furthermore, from the blade passing times T ₁₁ and T ₂₁ , (40),
The DC component of the first item of the equation (41) is removed. The method of removing this DC component is the same as that in step S3 of FIG. The equations at this time are the following equations (42) and (43). Then, the blade passage time T when all the DC components are removed
Frequency analysis is performed on the values of ₁₁ and T ₂₁ . The results of frequency analysis are averaged over all turbine blades. The state of this frequency analysis is shown in FIG. The order ratio is actually up to 0.5, and 0.5 to 1 are conjugate complex numbers. The average value of all turbine blades is the arithmetic average value of each order ratio (frequency) (the arithmetic average value of the same order ratio) as shown in FIG.

[0133]

[Equation 17]

[0134]

[Equation 18]

Subsequently, the spectrum calculated by the frequency analysis in the above step S56 is shown in FIG.
It develops like 1 (step S57 of FIG. 25). Then, based on this expanded spectrum, the above step S
The amplitude a and the phase φ of the vibration of each turbine blade 52 having the highest peak in the plural order ratios n calculated in 54 are searched (step S58 in FIG. 25).

Subsequently, the phase of each peak theoretically obtained from the following equations (44) and (45) is compared with the actual phase, that is, the phase of each turbine blade 52 obtained in step S58. , The order ratio n having the minimum error is set to the correct order ratio n (step S59 in FIG. 25).

[0137]

[Formula 19]

Next, the order ratio n of each peak obtained in step S59 is given as an initial value to the equations (36) and (37) which are equations of the nodal diameter mode method, and the curve is calculated by the nonlinear least squares method. By fitting, the most correct order ratio n is obtained (step S60 in FIG. 25).

Finally, if necessary, the above step S6
Correct nodal diamond number N using the order ratio n obtained by 0
d is calculated (step S61 in FIG. 25). That is, as in the case of step S55, the order ratio (n + Nd) obtained by the peak search by the frequency analysis of step S53 is performed.
Then, the value of Nd is obtained by subtracting the value of the order ratio n obtained in step S60. Then, based on the order ratio n obtained in step S60,
The amplitude a and the phase φ of the vibration of each blade are obtained. That is, the order ratio n obtained in step S60 is fed back to the equations (42) and (43), which are one-point equations, and the amplitude a and the phase φ of the vibration of each turbine blade 52 are nonlinearly minimized. Curve fitting or frequency analysis (FFT) by square method
(Step S61 in FIG. 25).

As described above, according to the fourth embodiment (nodal diameter mode method I), only two blade sensors 54-1 and 5-5 are provided.
With only 4-2, the vibration amplitude a of the turbine blade 52,
The phase φ and the frequency f _b can be obtained. For this reason,
The cost of measurement equipment such as blade sensors will be greatly reduced, and the time and cost required for sensor installation work and maintenance will be greatly reduced.

Further, in the blade vibration measuring system using this blade vibration measuring method, the blade sensors 54-1 and 54-2 are always operated.
Of the turbine blade 52 by obtaining the amplitude a, the phase φ, and the frequency f _b of the blade vibration based on the blade detection signals d _1N and d _2N of
Vibration monitoring can also be performed.

<Embodiment 5> ... Nodal Diameter Mode Method II In the above-described Embodiment 4 (nodal diameter mode method I), two blade sensors are used. However, the blade sensors have three equal pitches (intervals). When there is a point, the two-point method is used to focus on the fact that the same thing as in the above-mentioned Embodiment 4 (node diameter mode method I) is possible. is there. The nodal diameter mode method II will be specifically described below.

FIG. 32 is a sensor layout diagram of the blade vibration measuring system according to the fifth embodiment (node diameter mode method II) of the present invention,
FIG. 33 is a block diagram showing the configuration of the blade vibration measurement system according to the fifth embodiment (node diameter mode method II) of the present invention. 32 and 33, the same parts as those in FIGS. 20 and 22 are designated by the same reference numerals, and the duplicated description will be omitted.

As shown in FIGS. 32 and 33, the present embodiment
In form 5 (nodal diameter mode method II), three blade sensors 54-
1, 54-2 and 54-3 are used. These three wings
14-1, 14-2, 14-3, 14-4 have equal pitch
To be installed. That is, the wing sensors 54-1, 54-2, 54
-3 mounting angle (from the rotation pulse output position to each blade sensor)
Angle in S)₁, Sθ₂, Sθ₃Then, (Sθ
₃-Sθ₂) = (Sθ ₂-Sθ₁) = Δ.

The third blade sensor 54-3 is also a non-contact type sensor such as a proximity sensor like the other blade sensors, and the turbine blade 50 rotates in front of the blade sensor 54-3 by the rotation of the turbine rotor 50. The blade detection signal d _3N is output at the time when it passes (when the tip of the turbine blade faces the sensor). The blade detection signal d _3N are input to the clock 56-3.
The keyway detection signal (rotation pulse) e of the keyway sensor 15 is also input to the timepiece 56-4.

In the timepiece 16-4, when the first keyway detection signal e is input from the keyway sensor 55 as in the case of other timepieces, this is used as a trigger, and the subsequent blade sensors 54-
Based on the blade detection signals d _1N and d _2N of No. 3, all turbine blades 52
Measures the time T ₃ (N) for the blade to pass the blade sensor 54-3 in sequence, and outputs the blade passing time T ₃ (N) to the CPU 58.
That is, the blade passing time T ₃ (N) is a function of the blade number B _N , and assuming that the total number of turbine blades 52 is 60, for example, 60 points of measurement data can be obtained per revolution.

As will be described in detail later, the CPU 58 uses the clock 5
Blade passing time T input from 6-1, 56-2, 56-3
_1N, T_2N, T_3NRotation cycle input from the or clock 57
T_rVibration generated in each turbine blade 52 based on
Amplitude a, phase φ, frequency f _bCalculate (rotational order ratio n)
To do. Here, based on the flowchart of FIG.
The vibration amplitude a of the turbine blade 52 in the CPU 58,
Phase φ, frequency f_bThe calculation processing of will be described in detail.

First, the passage time T of all turbine blades 52
₁(N), T₂(N), T₃(N) is measured.
S71). The specific measuring method is as described above.
It In the case of all-round spelling mode vibration as described above,
Since the phase between the bin wings is given by a sine wave, here
Measured blade transit time T₁(N), T₂(N), T₃Formula (N)
It can be expressed by the following equations (46), (47), (48)
become. Also, the wing passage time T₁(N), T₂(N), T
₃(N) is measured data T for each turbine blade 52_1N, T
_2N, T_3NThen, the following (49), (50), (5
1) Expression (one-point method expression) is obtained (the above (1),
(Same as formulas (2) and (3)).

[0149]

[Equation 20]

[0150]

[Equation 21]

Then, the difference between the blade passing times T ₁ (N), T ₂ (N) and T ₃ (N) measured in step 71, that is, (T
₁ -T _2), calculates the _{(T 2 -T 3), (} T 1 -T 3) ( step 72 in FIG. 34). When these are expressed by equations, the following equations (51), (52) and (53) are obtained.

[0152]

[Equation 22]

Therefore, (T ₁ -T ₂ ) and (T ₂ -T ₃ )
As is clear from the comparison between the above equations (51) and (52), the phase difference between and becomes like the following equation (54).
The phase difference between (T ₁ -T ₂ ) and (T ₁ -T ₃ ) is expressed by the following equation (55), as is clear from the comparison between the equations (51) and (53). become that way.

[0154]

[Equation 23]

Therefore, the respective equations (54) and (55)
From this, the order ratio n can be obtained. Therefore, by analyzing this by a method similar to that of the above-described fourth embodiment (nodal diameter mode method I), that is, by executing the algorithm after step S52 in FIG. 25, the amplitude of vibration of each turbine blade 52 is increased. a, phase φ, and frequency f _b can be obtained (see FIG. 3).
4 step S63).

In the fourth embodiment (node diameter mode method I), the trend is removed when the DC component is removed in step S52 in FIG. 25. At this time, the fifth embodiment (node diameter mode) is used. In method II), the DC component can be removed by simply removing the average value. That is, the average value of T ₁ (N), T ₂ (N) and T ₃ (N) is simply obtained, and these average values are calculated as T ₁ (N), T ₂ (N) and T
_It may be removed from each of ₃ (N).

In addition, the order ratio n is (54), (55)
It may be obtained from any of the equations, or the average value of the values obtained by both equations may be used.

In the fourth embodiment (nodal diameter mode method I), the blade transit time T calculated by the one-point method is subjected to frequency analysis in step S56 of FIG. 25. In the node diameter mode method II), in the same step, the blade transit time T obtained by the two-point method is frequency-analyzed. That is, the blade passage times T _1N , T _2N , T of the turbine blades 52 represented by the above equations (51), (52), and (53) are represented.
Based on _3N , as in the case of the above-described first embodiment (optimization method I), the difference in blade passing time between each blade sensor is obtained, and
After removing the DC component (above (17), (18), (1
9) Equation)), and frequency analysis is performed.

Further, in the fourth embodiment (node diameter mode method I), in step S61 of FIG. 25, the order ratio n obtained in step S60 is fed back to the formula of the one-point method, and A curve fit or frequency analysis (F
FT), but in the present fifth embodiment (node diameter mode method II), in step S62, step S62 is performed.
The order ratio n obtained in 60 is fed back to the equation of the two-point method, and the amplitude a and the phase φ of the vibration of each turbine blade 52 are subjected to the curve fitting or the frequency analysis (FF) by the nonlinear least square method.
T).

As described above, according to the fifth embodiment (nodal diameter mode method II), the three blade sensors 54-1 and 54- are used.
The amplitude a, the phase φ, and the frequency f _b of the vibration of the turbine blade 52 can be obtained only by using 2, 54-3. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced. Furthermore, since the measurement time difference between the blade sensors is used, the blade sensors 54-1 and 5-5
Even if there is uniform shaking in 4-2 and 54-3,
Vibration measurement can be performed accurately.

In addition, in the blade vibration measuring system using this blade vibration measuring method, the blade sensors 54-1 and 54- are always operated.
It is also possible to monitor the vibration of the turbine blade 52 by determining the blade vibration amplitude a, phase φ, and frequency f _b based on the blade detection signals d _1N , d _2N , and d _3N of ₂ , 54-3.

In the above first to fifth embodiments, the case where the vibration of the turbine blade is measured has been described as an example. However, the present invention is not limited to this, and various rotating bodies such as compressors can be used. It can be applied to blade vibration measurement.

[0163]

As described above in detail with the embodiments of the invention, the blade vibration measuring method of the first invention is a method for measuring the vibration of a blade attached to a rotating body, and Based on the detection signals of the at least three non-contact blade sensors arranged, the first step (step S1 in FIG. 3) of measuring the blade passage time and the second step of calculating the difference between the blade passage times. Step (step S2 of FIG. 3) and third step (step S3 of FIG. 3) of removing a DC component (component generated between blade sensors) from the blade passage time difference.
The fourth step (step S4 of FIG. 3, step S4 of FIG. 3) for obtaining the amplitude A, the phase C, and the order ratio B of the blade vibration by frequency analysis of the blade passage time difference from which the DC component is removed by frequency analysis means. S5), and a fifth step (steps S6 and S7 in FIG. 3) of classifying blade vibration into an asynchronous component, a synchronous component, and a 0.5H component based on the order ratio B. Is (step S in FIG. 3)
8), a sixth step (step S8-1 in FIG. 4) of performing curve fitting on the expression of the phase C to obtain the order ratio n of the blade vibration and the phase φ, and the sixth step in the expression of the amplitude A.
A seventh step (step S8-2 in FIG. 4) of obtaining the amplitude a of the blade vibration using the order ratio n obtained in the step of
In the case of the synchronous component (step S9 in FIG. 3), the order ratio n
Is used as a parameter to perform a curve fit with respect to the equation of the two-point method, and an eighth step (step S9-1 in FIG. 5) of obtaining the amplitude a and the phase φ of the blade vibration, and the eighth step
A ninth step (step S9-2 in FIG. 5) of selecting the amplitude a and the phase φ by selecting the optimum condition from the order ratio n given as a parameter in the step of
In the case of the 0.5H component (step S10 in FIG. 3), a curve fitting is performed on the equation of the two-point method using the order ratio n as a parameter to obtain the amplitude a and the phase φ of the blade vibration. Step (step S10-1 in FIG. 6)
And an eleventh step (step S10-2 in FIG. 6) of selecting the amplitude a and the phase φ by selecting the optimum condition from the order ratio n given as a parameter in the tenth step, It is characterized by having.

Therefore, according to the blade vibration measuring method of the first aspect of the present invention, the amplitude, phase and order ratio (frequency) of the blade vibration can be obtained by using only three blade sensors. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced. Further, since the measurement time difference between the blade sensors is used, the vibration can be accurately measured even when the blade sensors are uniformly shaken.

Further, the blade vibration measuring method of the second invention is the first
In the blade vibration measuring method of the invention, the frequency analyzing means in the fourth step obtains the amplitude A, the phase C, and the order ratio B of the blade vibration by performing a Fourier transform on the blade passage time difference from which the DC component has been removed. With the twelfth step (step S4 in FIG. 3) and the blade passage time difference from which the DC component is removed, with the amplitude A, the phase C, and the order ratio B of the blade vibration obtained in the twelfth step as initial values, A thirteenth step (step S5 in FIG. 3) of performing curve fitting to obtain the amplitude A, the phase C, and the order ratio B of the blade vibration.

Therefore, according to the blade vibration measuring method of the second invention, the blade vibration measuring accuracy is further improved.

In addition, the blade vibration measuring method of the third invention is the same as the second method.
In the blade vibration measuring method of the invention, in the case of an asynchronous component, after the seventh step of the first invention, a curve fitting condition Z of the order ratio is created by using the order ratio B obtained in the thirteenth step. 14th step (step S8-3 in FIG. 3) and a fifteenth step (FIG. 3 in FIG. 3) for obtaining the amplitude a and the phase φ of the blade vibration by performing curve fitting based on the curve fitting condition Z. Step S8
-4), and the 16th step of selecting the optimum condition of the curve fitting condition Z from the result of the curve fitting, and selecting the order ratio n of the blade vibration, the amplitude a, and the phase φ (step S8- in FIG. 3). 5) and are included.

Therefore, according to the blade vibration measuring method of the third invention, the blade vibration measuring accuracy is further improved.

Further, the blade vibration measuring method of the fourth invention is the blade vibration measuring method of the first, second or third invention, in which the predicted value of the static test by numerical analysis or tapping is used when curve fitting is performed. It is characterized by being reflected as a boundary condition.

Therefore, according to the blade vibration measuring method of the fourth aspect of the present invention, the blade vibration measuring accuracy is further improved.

The blade vibration measuring method according to the fifth aspect of the invention is a method for measuring the vibration of a blade attached to a rotating body, wherein the outer periphery of the blade is replaced by the first to third steps of the first aspect of the invention. Seventeenth, which measures the blade transit time of each blade based on the detection signals of at least four non-contact blade sensors arranged in the pitch
(Step S21 in FIG. 14), the eighteenth step (step S22 in FIG. 14) for calculating the difference between blade passing times, and the nineteenth step for calculating the difference in blade passing time difference to remove the DC component. Step (step S2 in FIG. 14)
3), and the algorithm after the fourth step of the first invention is characterized by performing the algorithm of the first, second, third or fourth invention.

Therefore, according to the blade vibration measuring method of the fifth aspect of the present invention, the amplitude, phase, and order ratio (frequency) of the blade vibration can be obtained by using only four blade sensors. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced. Further, since the measurement time difference between the blade sensors is used, the vibration can be accurately measured even when the blade sensors are uniformly shaken.

The blade vibration measuring method of the sixth invention is a method for measuring the vibration of a blade attached to a rotating body, wherein the blade vibration measuring method is arranged on the outer periphery of the blade instead of the first and second steps of the first invention. 20th step (step S31 of FIG. 17) of measuring the passage time of the blade based on the detection signals of the at least two non-contact blade sensors, and from this blade passage time, the reference point of rotation is always used. The 21st step of removing the DC component by taking the time difference (step S3 of FIG. 17)
2) and, and the algorithm after the third step of the first invention is characterized by implementing the algorithm of the first, second, third or fourth invention.

Therefore, according to the blade vibration measuring method of the sixth aspect of the present invention, the amplitude, phase and order ratio (frequency) of the blade vibration can be obtained by using only two blade sensors. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced.

Further, the blade vibration measuring method of the seventh invention is a blade vibration measuring method when the blades attached to the rotating body vibrate in a coupled state, and at least three blades arranged on the outer periphery of the blade are arranged. A first step (step S51 in FIG. 25) of measuring the passage time of the blade based on the detection signal of the non-contact blade sensor at the point and a second step (FIG. 25) of removing the DC component from the passage time of the blade. 25) and the blade passage time from which the DC component has been removed, frequency analysis is performed to search for a peak value, and then a third step (step S53 in FIG. 25) of confirming the phase difference for each peak. )
And a fourth step (step S54 in FIG. 25) of calculating the order ratio n based on this phase difference, and the order ratio,
Combined with the results of the frequency analysis, the Nodal diamond number N
Fifth step of calculating d (step S5 of FIG. 25)
5), a sixth step (step S56 in FIG. 25) of frequency analysis of the passage time of each blade, and a seventh step (step of FIG. 25) of developing the spectrum calculated by this frequency analysis in consideration of folding back. S57) and the eighth step of searching the amplitude and phase of the vibration of each blade having the highest peak in the order ratio n calculated in the fourth step based on the developed spectrum (step S58 in FIG. 25).
And the phase of each peak theoretically obtained is compared with the actual phase obtained in the eighth step to obtain a correct order ratio n, ninth step 8 (step S59 in FIG. 25).
Then, the order ratio n of each peak obtained in the ninth step is given as an initial value to the equation of the nodal diameter mode method to perform curve fitting, thereby obtaining the most accurate order ratio n.
0 step (step S60 of FIG. 25) and this first
The eleventh step (FIG. 25) of obtaining the amplitude a and the phase φ of the vibration of each blade based on the order ratio n obtained in the step 0
And step S61).

Therefore, according to the blade vibration measuring method of the seventh aspect of the present invention, the amplitude, phase and order ratio (frequency) of the blade vibration can be obtained by using only three blade sensors. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced.

The blade vibration measuring method of the eighth invention is a method for measuring the vibration of a blade attached to a rotating body in a coupled state, wherein the outer periphery of the blade is replaced by the first step of the seventh invention. Based on the detection signals of at least three non-contact blade sensors arranged at equal pitches, the twelfth step of measuring the blade passage time (step S71 in FIG. 34) and the difference between the blade passage times are calculated. And a thirteenth step (step S72 in FIG. 34), and the algorithm after the second step of the seventh invention is characterized by implementing the algorithm of the seventh invention.

Therefore, according to the blade vibration measuring method of the eighth aspect of the present invention, the amplitude, phase and order ratio (frequency) of the blade vibration can be obtained by using only three blade sensors. Therefore, the cost of measurement equipment such as a blade sensor is significantly reduced, and the period and cost required for sensor installation work and maintenance are also significantly reduced. Further, since the measurement time difference between the blade sensors is used, the vibration can be accurately measured even when the blade sensors are uniformly shaken.

In addition, the blade vibration monitoring system of the ninth invention is
The blade vibration data measured by the blade vibration measuring method of the first, second, third, fourth, fifth, sixth, seventh, or eighth invention is transmitted to a monitoring device to monitor the blade vibration. It is characterized in that it is configured in.

Therefore, according to the blade vibration monitoring system of the ninth invention, the blade vibration is constantly monitored by obtaining the amplitude, phase and order ratio (frequency) of the blade vibration based on the blade detection signal of the blade sensor. It can be performed.

[Brief description of drawings]

FIG. 1 is a sensor layout diagram of a blade vibration measurement system according to a first embodiment (optimization method I) of the present invention.

FIG. 2 is a block diagram showing a configuration of a blade vibration measurement system according to a first embodiment (optimization method I) of the present invention.

FIG. 3 is a flowchart of a blade vibration measuring method according to the first embodiment (optimization method I) of the present invention.

FIG. 4 is a flowchart of a blade vibration measuring method according to the first embodiment (optimization method I) of the present invention.

FIG. 5 is a flowchart of a blade vibration measuring method according to the first embodiment (optimization method I) of the present invention.

FIG. 6 is a flowchart of a blade vibration measuring method according to the first embodiment (optimization method I) of the present invention.

FIG. 7 is an explanatory diagram of a blade passing time.

FIG. 8 is an explanatory diagram of a blade passing time.

FIG. 9 is an explanatory diagram of a blade vibration state.

FIG. 10 is an explanatory diagram showing an example of frequency analysis.

FIG. 11 is an explanatory diagram showing a method of correcting a phase C.

FIG. 12 is a sensor layout diagram of a blade vibration measurement system according to a second embodiment (optimization method II) of the present invention.

FIG. 13 is a block diagram showing a configuration of a blade vibration measurement system according to a second embodiment (optimization method II) of the present invention.

FIG. 14 is a flowchart of a blade vibration measuring method according to a second embodiment (optimization method II) of the present invention.

FIG. 15 is a sensor layout diagram of a blade vibration measurement system according to a third embodiment (optimization method III) of the present invention.

FIG. 16 is a block diagram showing a configuration of a blade vibration measurement system according to a third embodiment (optimization method III) of the present invention.

FIG. 17 is a flowchart of a blade vibration measuring method according to a third embodiment (optimization method III) of the present invention.

FIG. 18 is an explanatory diagram of blade passage time.

FIG. 19 is an explanatory diagram of the reason why at least two blade sensors are required.

FIG. 20 is a fourth embodiment of the present invention (node diameter mode method I).
3 is a sensor layout diagram of the blade vibration measurement system according to FIG.

FIG. 21 is an explanatory diagram showing a structural example of a turbine blade.

FIG. 22 is a fourth embodiment of the present invention (node diameter mode method I).
2 is a block diagram showing the configuration of a blade vibration measurement system according to FIG.

FIG. 23 is an explanatory diagram of a blade vibration state.

FIG. 24 is an explanatory diagram of a blade vibration state.

FIG. 25 is a fourth embodiment of the present invention (nodal diameter mode method I).
5 is a flowchart of a blade vibration measuring method according to the present invention.

FIG. 26 is an explanatory diagram of a DC component removing method.

FIG. 27 is an explanatory diagram of a DC component removing method.

FIG. 28 is an explanatory diagram of frequency analysis.

FIG. 29 is an explanatory diagram of frequency analysis.

FIG. 30 is an explanatory diagram of a method of averaging all wings.

FIG. 31 is an explanatory diagram showing an example of expansion of a spectrum.

FIG. 32 is a fifth embodiment of the present invention (node diameter mode method II).
3 is a sensor layout diagram of the blade vibration measurement system according to FIG.

FIG. 33 is a fifth embodiment of the present invention (node diameter mode method II).
2 is a block diagram showing the configuration of a blade vibration measurement system according to FIG.

FIG. 34 is a fifth embodiment of the present invention (node diameter mode method II).
5 is a flowchart of a blade vibration measuring method according to the present invention.

[Explanation of symbols]

10 turbine rotor 11 rotation axis 12 turbine blades 13 keyway 14-1, 14-2, 14-3, 14-4 Wing sensor 15 key groove sensor 16-1, 16-2, 16-3, 16-4 clock 17 clock 18 CPU 19 transmitter 20 remote monitoring equipment 21 receiver 22 CPU 23 Display 50 turbine rotor 51 rotation axis 52 turbine blades 53 keyway 54-1, 54-2, 54-3 Wing sensor 55 key groove sensor 56-1, 56-2, 56-3 Clock 57 clock 58 CPU 59 transmitter 60 remote monitoring equipment 61 receiver 62 CPU 63 indicator

Continuation of front page (56) Reference JP-A-2-23208 (JP, A) JP-A-4-246207 (JP, A) JP-A-5-65803 (JP, A) JP-A-6-201506 (JP , A) JP 7-43206 (JP, A) JP 7-128133 (JP, A) JP 61-101605 (JP, A) JP 63-229333 (JP, A) JP 2001 -165089 (JP, A) Patent 3129406 (JP, B2) Patent 2935503 (JP, B2) Patent 3095270 (JP, B2) United States Patent 5511426 (US, A) United States Patent 5974882 (US, A) United Kingdom Patent Application Publication 2344177 ( (GB, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G01H 17/00 F01D 25/00 G01M 19/00

Claims (9)

(57) [Claims] 1. A method for measuring the vibration of a blade attached to a rotating body, which comprises measuring the passage time of the blade based on the detection signals of at least three non-contact blade sensors arranged on the outer periphery of the blade. Step 1, a second step of calculating the difference between the blade passing times, a third step of removing a DC component from the blade passing time difference, and a frequency analysis of the blade passing time difference from which the DC component is removed. The frequency analysis is performed by means to measure the amplitude A and the phase C of the blade vibration.
And a fourth step of obtaining the order ratio B, and a fifth step of classifying the blade vibration into an asynchronous component, a synchronous component, and a 0.5H component based on the order ratio B, and in the case of the asynchronous component, , A sixth step of curve-fitting the equation of the phase C to obtain the order ratio n and phase φ of the blade vibration, and an equation of the amplitude A to the order ratio n obtained in the sixth step.
The 7th step to find the amplitude a of the blade vibration using, and in the case of the synchronous component, by performing a curve fit to the equation of the 2-point method using the order ratio n as a parameter, the amplitude a and the phase of the blade vibration Eighth step of obtaining φ and the order ratio n given as a parameter in this eighth step
By selecting the optimum condition from among these, the ninth step of selecting the amplitude a and the phase φ, and in the case of the 0.5H component, using the order ratio n as a parameter, the curve fit to the equation of the two-point method Is performed, and by selecting the optimum condition from the tenth step of obtaining the amplitude a and the phase φ of the blade vibration, and the order ratio n given as a parameter in this tenth step, the amplitude a and the phase are selected. An wing vibration measuring method comprising: an eleventh step of selecting φ. 2. The blade vibration measuring method according to claim 1, wherein the frequency analyzing means in the fourth step performs Fourier transform on the blade passing time difference from which the direct current component is removed, and the amplitude A of the blade vibration. , The phase C, the twelfth step of obtaining the order ratio B, and the blade passage time difference from which the DC component is removed, with the blade vibration amplitude A, the phase C, and the order ratio B obtained in the twelfth step as initial values. On the other hand, a blade vibration measuring method comprising: a curve fitting to obtain an amplitude A, a phase C, and an order ratio B of the blade vibration. 3. The blade vibration measuring method according to claim 2, wherein in the case of an asynchronous component, after the seventh step of claim 1, the order ratio B obtained in the thirteenth step is used. A fourteenth step of creating a curve fit condition Z of the order ratio, and a fifteenth step of obtaining the amplitude a and the phase φ of the blade vibration by performing a curve fit based on the curve fit condition Z, and this curve A blade vibration measuring method comprising: a 16th step of selecting the optimum condition of the curve fitting condition Z from the result of the fitting, and selecting the order ratio n of the blade vibration, the amplitude a, and the phase φ. 4. The blade vibration measuring method according to claim 1, 2, or 3, characterized in that when curve fitting is performed, a predicted value of a static test by numerical analysis or tapping is reflected as a boundary condition. Wing vibration measurement method. 5. A method for measuring vibration of a blade attached to a rotating body, wherein at least four points of non-contact arranged on the outer periphery of the blade at equal pitches, in place of the first to third steps of claim 1. Based on the detection signal of the blade sensor, the seventeenth step of measuring the blade transit time of each blade, the eighteenth step of calculating the difference in blade transit time, and the difference in the blade transit time difference are calculated. A blade vibration measurement, which comprises a nineteenth step of removing a DC component, and the algorithm after the fourth step of claim 1 implements the algorithm of claim 1, 2, 3 or 4. Method. 6. A method for measuring vibration of a blade attached to a rotating body, wherein at least two non-contact blade sensors arranged on the outer circumference of the blade are replaced with the first and second steps of claim 1. 20th which measures the passage time of the wing based on the detection signal of
And a 21st step of removing a DC component by always taking a time difference from a reference point of rotation from the blade passage time, and the algorithm after the 3rd step of claim 1 A blade vibration measuring method, characterized in that the algorithm of claim 1, 2, 3 or 4 is implemented. 7. A blade vibration measuring method when blades attached to a rotating body vibrate in a coupled state, the detection signals of at least three non-contact blade sensors arranged on the outer periphery of the blade. The first step of measuring the blade transit time, the second step of removing the DC component from the blade transit time, and the frequency analysis of the blade transit time from which the DC component was removed The third step of searching the peak value and then confirming the phase difference for each peak, the fourth step of calculating the order ratio n based on this phase difference, the order ratio, and the result of the frequency analysis In addition, the fifth step of calculating the Nodal diamond number Nd, the sixth step of frequency-analyzing the passage time of each blade, and the seventh step of developing the spectrum calculated by this frequency analysis in consideration of folding back. And an eighth step of searching the amplitude and phase of the vibration of each blade having the highest peak in the order ratio n calculated in the fourth step based on the developed spectrum, and the phase of each peak theoretically obtained. And the actual phase obtained in the eighth step to obtain a correct order ratio n, and a ninth step of obtaining the order ratio n of each peak obtained in the ninth step is calculated according to the node diameter mode method. By giving a curve fit to the equation as an initial value, the tenth step of obtaining the most accurate order ratio n, and the amplitude a of the vibration of each blade based on the order ratio n obtained in this tenth step A blade vibration measuring method comprising: an eleventh step of obtaining a phase φ; 8. A method for measuring vibration of a blade attached to a rotating body in a coupled state, wherein at least three non-contact points arranged on the outer periphery of the blade at equal pitches are used instead of the first step of claim 7. The twelfth step of measuring the passage time of the blade based on the detection signal of the contact blade sensor, and the thirteenth step of calculating the difference between the blade passage times, the second step of claim 7. The blade vibration measuring method, wherein the algorithm after the step implements the algorithm of claim 7. 9. A structure for transmitting blade vibration data measured by the blade vibration measuring method according to claim 1, 2, 3, 4, 5, 6, 7 or 8 to a monitoring device to monitor blade vibration. A blade vibration monitoring system characterized by the above.