JPS6044613B2 - Diagnostic device for rotating machines - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a device for diagnosing abnormalities in rotating machinery, particularly an FF
The present invention relates to a movable diagnostic device for a rotating machine that incorporates a T and is capable of diagnosing the presence or absence of an abnormality in a rotating machine by performing spectrum analysis of the vibration of the rotating machine on site.

Predictive maintenance equipment diagnosis technology has been recognized for its effectiveness in effectively proceeding with equipment maintenance work, and is being introduced to on-site maintenance work and is becoming effective.

The vibrations of rotating machines, reciprocating machines, etc., which account for the majority of these equipment, are detected, and if the overall amplitude becomes abnormally large compared to the normal value, it can be determined that some kind of abnormality has occurred. In other words, the presence or absence of an abnormality can be determined by the magnitude of the overall value of vibration amplitude. Furthermore, in rotating machinery, it is well known that by analyzing the frequency spectrum from the overall value of the detected vibration and calculating the amplitude of each frequency spectrum, it is possible to specifically determine the cause of the abnormality in the target to be diagnosed. is being executed.

By the way, conventionally, abnormality diagnosis of this type of rotating machinery involves bringing a vibration recorder to the site, recording vibration data of the rotating machine to be diagnosed, and then taking the data back and analyzing it using a large spectrum analyzer. Because the analysis method was adopted, real-time analysis was not possible, which was extremely inconvenient. For this reason, FFT (FastFourierTransform) using a digital arithmetic unit is used.
A diagnostic device has been proposed that incorporates a frequency spectrum analyzer of the m) method and can be brought to the field to diagnose the presence or absence of abnormalities in rotating machinery in real time.

In other words, this diagnostic device has a weight of about 10.9 to 30.9, and when it is brought to a maintenance site, it performs real-time frequency analysis on the rotating machine side and determines the rotational speed of the object to be diagnosed as the basic cycle. When an amplitude spectrum occurs at a specific frequency of vibration, the abnormal location and cause of the abnormality in the rotating machine is determined from the relationship table between the mechanical vibration spectrum and the abnormal location or cause, which is theoretically determined in advance. That's what I do.

When equipment maintenance personnel carry out these judgment tasks, currently, the frequency spectrum distribution calculated by a diagnostic device based on frequency analysis and the magnitude of the amplitude of the line spectrum for each frequency are displayed on a CRT display. (In this case, the spectral amplitude for each frequency can be determined by moving the cursor, displaying the spectral amplitude corresponding to the specified frequency in numbers, and reading this to confirm the amplitude of each line spectral value. There is.

) Or, by recording the frequency spectrum distribution calculated using a pen-writing X-Y recorder so that the horizontal axis represents frequency and the vertical axis represents amplitude, and reading this frequency spectrum value, it is possible to determine the target to be diagnosed. Determines whether there is an abnormality in the rotating machine, the location of the abnormality, and its cause. Now, if we focus on the FFT method that uses a digital arithmetic unit to calculate these frequency spectrum distributions, as is well known,
Calculating the frequency spectrum distribution using the FFT method has the disadvantage that it has only a finite resolution due to the nature of digital calculations. This resolution is determined by the length of the data calculation register regarding the amplitude of the calculated frequency spectrum. For example, if the register length is 8 bits, the full scale input is 1×256, and if the register length is 10 bits, data having a resolution of 1×024 can be obtained. On the other hand, when calculating a frequency spectrum by performing FFT calculation, the frequency resolution must also be a finite value. Ru.

The frequency resolution is determined by the sampling time interval (analog-digital conversion interval) and the number of input data, as is well known in FFT calculation. That is, the sampling time interval is Δ
t and the number of input data is N=2m, the maximum analysis frequency.゜The large value FM becomes FM=112Δt, and the resolution Δf of the analysis frequency becomes.

In other words, analysis results of the frequency spectrum up to FM appear every Δf.

Therefore, the number of frequency spectra is NI2.
The vibration value input detected here is nΔf (n is an integer n=0,
1, 2, ... NI2-1), i.e., f(t) = Asi
When n2πFmt, the result of frequency spectrum analysis of this input by FFT is not a single line spectrum, that is, a single frequency line spectrum, but is expressed as a mixture of various frequency spectra.

For example, if the number of input data is 1024 and the sampling time interval is 9.765 x 10-4 seconds, the maximum value of the analysis frequency will be , and the analyzed frequency spectrum will be displayed every 1 Hz from 1-Hz to 512 Hz. Can be done.

The amplitude value input detected here is f(t)=Asjn2πF
mt, if Fm is 50.0Hz, the spectrum will be a line spectrum, but if Fm is 49.7Hz49.
8Hz49.9Hz50.1Hz50.2Hz...
If it is 50.6 Hz, etc., it will not be a line spectrum, but will have a spectral distribution near the input frequency, that is, it will be displayed as if a mixed input of harmonic oscillations were spectral distributed, rather than a single harmonic oscillation.

This means that when a vibration value having a vibration period other than an integral multiple of the resolution of the analysis frequency is input, its amplitude cannot be accurately known.

For this reason, when bringing a frequency spectrum analysis type diagnostic device to the site and performing real-time abnormality diagnosis of the rotating machine on the rotating machine side, it is not only difficult to obtain the absolute value of the vibration width, but also to compare it with previous diagnostic data. However, trend management based on the above method also has an essential drawback in that it becomes difficult to perform it in real time on site. Therefore, the present invention has been made to solve the above-mentioned problems in spectrum analysis in a movable rotating machine diagnostic device that employs the FFT frequency spectrum analysis method, and the analysis frequency is not an integer multiple of the analysis frequency. The basic object of the present invention is to provide a diagnostic device for a rotating machine that can provide an accurate spectrum amplitude value even when a vibration input has a frequency. Therefore, in the present invention, spectrum analysis is performed using the FFT method, the magnitude of the frequency spectrum values obtained as a result is determined, the upper amplitude value including the maximum value is selected, for example, the amplitude value of one circle, and the selected amplitude The basic feature is that the sum of squares and square root calculation is performed by selecting amplitude values, for example, 1 mile before and after the central value.

Hereinafter, the present invention will be described in more detail based on illustrated embodiments.

In the drawing, 1 is a vibration detection device that is attached to a rotating machine to be diagnosed and converts mechanical vibration characteristics into an electric signal, and 2 is a vibration detection device that combines a high frequency band cutoff filter 2a and a low frequency band cutoff filter 2b, and is used to diagnose abnormalities in the rotating machine. A bandpass filter cuts out unnecessary frequency components; 3 is an analog-digital converter that converts the analog electrical signal that has passed through the bandpass filter 2 into a digital signal; 4 is an analog-to-digital converter that converts the analog electrical signal passed through the bandpass filter 2 into a digital signal A digital frequency spectrum analysis calculator 5 called FFT that analyzes the frequency spectrum of a digital signal determines the magnitude of the spectrum amplitude obtained by the digital frequency spectrum analysis calculator 4, and calculates the magnitude of the spectrum amplitude in descending order of the magnitude, for example, 1 yen from the highest value. A spectral amplitude size discriminator 6 selects the amplitude value of the spectral amplitude size discriminator 5, and a spectral amplitude size discriminator 6 selects a decomposition frequency Δf (for example, 1 Hz) centered on the frequency FO that gives the maximum amplitude value obtained by the spectral amplitude size discriminator 5.
7 is a spectral amplitude magnitude discriminator 5 and a digital printer that prints out the calculation results of the sum of square root calculator 6 as needed;
8 is a control circuit for controlling the operation of these circuits.

As described above, when there is an abnormality in the rotating machine, the frequency spectrum exhibits a spectrum having a significant amplitude at a frequency corresponding to the cause of the abnormality.

The spectral amplitude size discriminator 5 determines significant amplitude values from the amplitude values output by the digital frequency spectrum analysis calculator 4 for each frequency that is an integral multiple of the decomposition frequency set for the calculator 4. The sum-of-squares-square-root calculation unit 6 performs the necessary sum-of-square-square-root calculation using the large amplitude value selected by the spectral amplitude size discriminator 5 as a guide.

As mentioned above, most of the remarkable spectral amplitudes corresponding to the causes of abnormalities occur at frequencies equal to or equal to an integral multiple of the fundamental frequency FO of the amplitude generated by the rotating machine. Digital converter 3
and digital frequency spectrum analysis calculator 4
If it does not match an integer multiple of the frequency resolution Δf determined by It is calculated as a combination of the line spectra of

However, the true amplitude value cannot be easily read with such a combination of many line spectra. Therefore, in the present invention, when the above-mentioned sum-of-squares square root is executed from the frequency spectra of adjacent front and rear, for example, w sets including the selected frequency, this value is the largest value among the frequency spectra of vibrations generated by the rotating machine. We take advantage of the fact that it can be expressed approximately. That is, if the maximum frequency spectrum is printed as a mixture of frequency spectra that are shifted by the resolution frequency determined by FFT calculation in the vicinity of the fundamental frequency, it is determined that the input signal is not an integral multiple of the resolution frequency.

In this case, as shown in the following table based on experimental results and theoretical formula confirmation, approximately 98%
The vibration frequency spectrum amplitude of the rotating machine to be diagnosed can be obtained using the above approximations.

Through these processes, even if the vibration frequency determined from the rotational speed of the rotating machine to be diagnosed does not match the integral multiple of the FFT resolution frequency, the vibration spectrum of the rotating machine to be diagnosed can be obtained in the form of a line spectrum. In other words, the true amplitude of vibration input can be obtained with an accuracy that does not pose a problem in practical use. If the maximum frequency in the frequency spectrum of vibrations generated by a rotating machine matches an integral multiple of the resolution frequency determined by the analog-to-digital converter 3 and the digital calculation frequency spectrum analysis calculator 4, it is a digital calculation frequency spectrum. When the frequency spectrum is calculated by the analysis calculator 4, it has already been processed into a line spectrum state, and even if the sum of squares of the frequency spectrum of one adjacent circle including that frequency is squared, the amplitude of the adjacent frequency spectrum will not be the same. The value is almost zero, and after all, even in such a case, it is possible to calculate a line spectrum with accurate amplitude. After calculating the maximum frequency spectrum of the vibrations generated by the rotating machine into an accurate line spectrum, we extract the frequency spectrum with the largest amplitude from the remaining frequency spectra, excluding the one frequency spectrum used for this calculation. The frequency is selected by a spectrum amplitude size discriminator 5.

Next, in the same way as above, the amplitudes of the frequency spectra of one adjacent circle including the selected frequency are square rooted.

The amplitude obtained in this way is the second largest frequency spectrum amplitude in the frequency spectrum of the vibration generated by the rotating machine. It is calculated as the amplitude of the vector. Of course, even if the frequency of the frequency spectrum matches an integral multiple of the frequency resolution determined by the above-mentioned calculator, an accurate amplitude can be obtained even if the processing by the sum-of-squares square root calculator is executed as described above.

If the above-mentioned method is again used to remove the m sets of frequency spectra used in the above calculation (excluding 2 circles in total) and calculate the remaining frequency spectra, the third largest amplitude value of the frequency spectrum will be obtained.

These sequential operations are instructed and executed by a sequence memory built into the control circuit 8.

In addition, in order to effectively proceed with the predictive maintenance equipment diagnosis of the rotating machine to be diagnosed, the effectiveness will be increased by continuing to carry out periodic inspections and periodic diagnosis and appropriately managing trends. According to the present invention, even if the rotational speed of the rotating machine to be diagnosed changes slightly and the distribution of the analyzed frequency spectrum changes, reliable spectrum amplitude values can be obtained reliably, making it easy for the examiner to make judgments. Based on this, the above-mentioned inconvenience can be solved because it can be processed in real time on site. The portable rotating machine diagnostic device implementing the present invention is
The amplitude spectrum at each frequency of the vibration value input detected in real time at the equipment maintenance site can be calculated.
This is particularly effective in managing vibration spectrum trends during on-site equipment diagnosis work.

[Brief explanation of the drawing]

The drawing is a block circuit diagram of a rotating machine diagnostic device according to the present invention. 1... Vibration detection device, 2... Bandpass filter, 3... Analog-digital converter, 4...
...Digital frequency spectrum analysis calculator, 5...
...Spectrum amplitude size discriminator, 6...Sum of squares square root calculator.

Claims (1)

[Claims] 1. Vibration detection means that detects the vibration of a rotating machine as an electrical signal, a bandpass filter that can adjust the passband in response to the reference vibration of the rotating machine, and a converter that converts the analog signal that has passed through the bandpass filter into a digital signal. a digital frequency calculator that calculates the frequency spectrum distribution from the input digital signal; a spectrum amplitude size discriminator that selects the upper amplitude value from the calculated frequency spectrum; Centering on the frequency that gives the highest amplitude value selected by the digital frequency spectrum analysis calculator, multiple sets of spectral amplitudes before and after each frequency interval given as the resolution of the digital frequency spectrum analysis calculator are squared and summed. A diagnostic device for a rotating machine, characterized by comprising a sum-of-squares square root calculator that performs square root calculation, and a recording device that records the calculation results.