JPS5853727B2 - Plate shape detection method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a plate shape detection method and apparatus that accurately and quickly obtain the shape of a plate in the width direction.

Automation of rolling mills has progressed in recent years, and most of the rolling mills that are newly built today are equipped with AGC systems and computers.

Although the flatness detection technology for the shape in the lengthwise direction of the board has been considerably advanced, the technology to automatically obtain the flatness of the shape in the width direction of the board has not yet been established.

One of the reasons why the technology for detecting flat dust in the width direction of the plate has not been perfected is that
It is a well-known fact that there is no device for detecting flat dust that is of high practical value.

Flat dust particles are flat when the difference in relative elongation of the plate in the amplitude direction is smaller. need to be detected.

Conventional shape detection methods include applying force to a board to detect sag, and looking at the tension distribution based on the difference in sag.Also, arranging split rolls and installing a load cell for each, pressing the roll against the board. Detect the difference in force with a load cell,
There are various methods such as methods of determining tension distribution, methods of projecting light and observing its reflection and fluctuation of a virtual image, and methods of determining the stress of a plate by the difference in its magnetic permeability.

However, in the conventional method, the state of tension distribution can be directly determined from the difference in the output of sensors arranged in the width direction of each sheet. Due to the effects of expansion and the winding thickness of the coil, the statically detected difference signal in the width direction may disappear into noise, making it impossible to perform highly accurate detection.

The present invention was made with the aim of eliminating the above-mentioned drawbacks of the conventional method, and detects the displacement amount signal induced by the tension distribution of the plate material by a displacement meter arranged in the width direction of the plate material. is regarded as the number of periodic strokes for each board width, and the correlation between the number of periodic strokes and the cosine wave is taken. As a result, it is possible to remove irregular signal components such as mechanical vibrations and electrical noise. .

Embodiments of the present invention will be described below with reference to the drawings.

In Figs. 1 to 3, 1 is a rolling machine, 2 is a plate whose shape is to be detected, 3 is n pads arranged in the width direction of the plate, 4 is a displacement meter installed in the pad 3, and 5 is a plate material whose shape is to be detected. Flow of fluid, 6 is a valve, 7 is a high-pressure fluid, 8 is a plate 2 displaced by the pressure of the fluid 7
10 is a timing circuit for turning on and off the valve 6, 11 is a sampler for collecting signals detected by each displacement meter 4, and 12 is sent from the sampler 11. Memory for storing data 13 is a read-only memo IJ-114 is a multiplier, 15 is an average circuit, 16 is a memo for storing data sent from the average circuit 15 U-117 is an amplitude phase adjustment circuit, 18 is a synthesis circuit , 19 is a CRT monitor, 20 is a timing setting signal, 21 is a read pulse, and 22 is a shape signal.

Now, when the valve-on signal from the timing circuit 10 is sent to the valve 6, the valve 6 opens, a fluid flow 5 occurs, pressure is generated in the pad 3, and the plate 2 is moved as shown in 8 in FIG. Displace.

This displacement of the plate material 2 is determined by the tension of the plate material 2, and is detected by the displacement meter 4.

That is, since the plate material 2 is displaced in accordance with the tension, the tension can be determined by measuring the displacement.

However, as the plate progresses, this displacement changes due to the influence of mechanical vibration, making it difficult to detect a value corresponding to the true tension.

However, the vibrations occur irregularly, while the deflection does not change significantly within a given period of time.

Therefore, by repeatedly measuring the deflection after displacement and correlating the signals, random mechanical noise can be eliminated and a signal representing only the tension of the plate material 2 can be obtained.

Next, the calculation method will be explained in more detail.

In FIG. 3, from the output signal of the timing circuit 10,
When the valve 6 is turned on, pressure is generated on the pad 3, as described above, and the plate 2 is displaced upward.

The n displacement meters 4 arranged in the width direction of the plate detect the amount of displacement that changes depending on the tension of the plate 2, and the detected displacement is sent as a signal to the memory 12 via the sampler 11, and is stored in the memory 12. Stored at n addresses.

Timing circuit 10 turns valve 6 on and off intermittently;
When turned on, the sampler 11 located between the displacement meter 4 and the memory 12 is in an open state.

Also, when the valve is off, the sampler 11 is closed.

While the valve 6 is off (closed), the values stored in the memory 12 and the values 1 to 1 in the read-only memory 13 are
A correlation with one period of the cosine wave stored at address n is taken.

That is, the value for one cycle of the cosine wave at address n of the read-only memory 13 is read out cyclically by the read pulse 21 and supplied to the multiplier 14, and the multiplier 14 reads out the value for one cycle of the cosine wave at address n and the memory 12. The n multiplied signals obtained by the multiplication are sent to the averaging circuit 15, where they are added together and divided by n, and the averaged be converted into

In this way, the correlation stroke value at one point in the board width direction at a certain time (point in time) is determined.

Then, if the above-mentioned certain point is time n1, the correlation function value at one point in the board width direction at time n1 is the memory 16
is stored at address n1.

Next, the cosine wave values in the read-only memory 13 are shifted by one and read out by the read signal 21, and the respective multiplications are combined and averaged, and then stored in the memory 16 at address nl+1.

This becomes the correlation value at time n1+1.

Similarly, the values of the number of correlated strokes corresponding to one cycle are stored in addresses 1 to n of the memory 16.

Since the value stored in the memory 16 corresponds to one harmonic of the periodic component included in the original displacement signal, this is compensated by the amplitude phase adjustment circuit 17 and the original signal is regenerated.
Input to the synthesis circuit 18.

The same thing is done for the harmonics necessary to approximate the board shape, and they are superimposed in the synthesis circuit 18 and displayed on the CRT monitor 1.
By displaying with 9, a plate shape can be obtained.

These series of calculations are performed until the valve 6 is next turned on.

Subsequently, the next time the valve 6 is turned on, the memory 12
The content of is rewritten, and the board shape is newly calculated by using the revised signal of the memory 12.

By taking the correlation in this way, irregular noise caused by vibration of the plate, etc. is completely removed, and accurate information on the plate shape can be obtained.

Next, the theory on which the above calculation method is based will be explained using mathematical formulas.

The gap displacement gauge arranged in the board width direction detects the displacement of the board in the vertical direction on which noise is superimposed, as shown in FIG.

Using this signal, the following measures are taken to detect only the displacement signal containing plate shape information.

As the plate material 2 advances, each time point t1 shown in FIG. t2
．． t3. A displacement signal with superimposed noise is detected every t4..., but when these are arranged on the horizontal axis as shown in Fig. 6, the plate shape changes rapidly as the plate 2 advances. Therefore, a periodic waveform whose plate width length corresponds to one cycle can be obtained.

That is, the displacement waveform depending on the plate shape forms a periodic waveform in which the period T is assigned to the amplitude length.

Therefore, the periodic waveform and COS (C05
(Xk=1.2.-・-n; m=L2.・-0
0) [where n is the number of gap displacement meters in the plate width direction], the m-th harmonic component of the periodic waveform can be detected.

Therefore, by synthesizing the first harmonic to the m-th harmonic, an approximate waveform of the noise-removed displacement signal up to the m-th harmonic can be obtained using a Fourier series.

Since the plate shape exhibits a monotonous change in the width direction, it is not necessary for m to take a very large value, and it is thought that m may be of the order of three at most.

The noise-superimposed displacement waveform in the width direction obtained at each time point ti (i-1, 2, 3, . . . ) can be expressed as the sum of Fourier series representation and Gaussian noise using the following equation.

get.

As can be seen from the formula (611), each obtained correlation stroke number is (
1) It is similar to each harmonic of the alternating current component of equation.

Therefore, by doubling the amplitude and compensating the phase for the result of calculating the correlation, and superimposing them, the alternating current component of equation (1) is obtained.

This represents the shape of the plate at a certain point in time.

In reality, as described above, this calculation is performed digitally using the outputs of n gap displacement meters arranged in the width direction of the plate.

In the calculation method described above, calculations can be made based only on the detection signal of one wavelength well at a certain point in time, but in order to improve accuracy, calculations using several wavelengths can be performed to lengthen the averaging time. Alternatively, the results calculated for each wavelength may be averaged for several wavelengths.

In the embodiments of the present invention, a case has been described in which force is applied intermittently to the pad using a valve, but it is also possible to apply force to the pad continuously, and this Of course, if it is a seed signal, it can be made into an accurate shape signal by the same calculation process as described above, and various other changes can be made without departing from the gist of the vehicle invention.

Since the plate shape detection method and device of the present invention have the above-described configuration, (1) Obtains a true signal of the plate shape from a signal buried in noise, which is a common drawback of conventional shape detection sensors. This problem can be solved by introducing a statistical method, and as a result, a true shape signal can be obtained easily and reliably.

(II) Conventional shape detection sensors often process signals in the traveling direction of the plate and then replace them in the width direction to detect the shape, which takes time to detect. According to this means, since signal processing is performed by focusing on the board width direction from the beginning, the detection time is shortened.

Various excellent effects can be achieved.

[Brief explanation of the drawing]

FIG. 1 is an explanatory perspective view showing a state in which a plurality of pads are arranged in the board width direction in the board shape detection method and device of the present invention, and FIG. 2 is an explanatory diagram of the board shape detection method and device of the present invention. , FIG. 3 is an explanatory diagram of the plate shape detection method and apparatus of the present invention, and FIG. 4 is an explanatory diagram of the principle of the plate shape detection method and apparatus of the present invention. 5 is an explanatory diagram of the principle of the plate shape detection method and apparatus of the present invention, and explanatory diagrams at each point in the width direction of the plate, and FIG. 6 are explanatory diagrams of the principle of the plate shape detection method and apparatus of the present invention. It is an explanatory diagram of the principle,
FIG. 2 is an explanatory diagram in which detection signals in the board width direction at each time point are arranged on the horizontal axis. 1 is a rolling machine, 2 is a plate material, 3 is a pad, 4 is a displacement meter, 6 is a valve, 10 is a timing circuit, 11 is a sampler, 12 is a memory, 13 is a read-only memo note, 14 is a multiplier,
15 is an average circuit, 16 is a memory, 1T is an amplitude and phase adjustment circuit, and 19 is a CRT monitor. 18 is a joint time

Claims (1)

[Claims] 1. The signal of the amount of displacement induced by the tension distribution of the plate material is detected by a displacement shape arranged in the width direction of the plate, and the signal is regarded as the number of periodic strokes for each width of the plate, and the number of periodic strokes and the cosine wave are A plate shape detection method characterized by taking a correlation with one period of . 2 Pads that are arranged in the width direction of the plate and apply tension to the plate using fluid pressure, a displacement meter that detects the amount of displacement of the plate, a memory that holds a signal of the amount of displacement detected by the displacement meter, and a signal that is sent from the memory. A multiplier that multiplies the received signal by the signal of the number of strokes of the cosine wave for one period sent from the read-only memory, another memory that holds the averaged signal obtained by multiplication, and a board. A plate shape detection device characterized by being provided with a synthesis circuit that combines amplitude and phase in the width direction and then synthesizes them.