JPH0346056B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a mechanical equipment abnormality diagnosing device for diagnosing abnormalities in industrial mechanical equipment.

Industrial machinery and equipment are responsible for product development, shipping inspection,
By measuring vibration at each stage of installation inspection, maintenance management, and maintenance, you can keep better machines operating in better condition.

In particular, maintenance management of machines has become important as machines have become larger and more sophisticated. It is necessary to manage the condition of machines on a daily basis, to detect abnormalities as early as possible, and to stop and repair them at the time when the overall impact is least. Therefore, there is a need for equipment abnormality diagnosis equipment as a tool for saving resources and reducing costs.

[Conventional technology and its problems to be solved]

Conventional devices of this type include small portable devices and large devices using large computers. In the former method, the signal obtained from the mechanical equipment to be diagnosed by vibration pickup is divided into each component using low, medium, and high frequency filters, and by observing the temporal level changes of each component, the normality of the mechanical equipment can be determined. Although it is used to diagnose different types of abnormalities, there is a problem in that it is difficult to determine what is causing the abnormality unless the mechanical equipment is inspected manually.

On the other hand, the latter performs frequency analysis over a wide frequency range of the vibration signal obtained by pickup to obtain a spectrum at each frequency, and then compares the entire spectrum with a pre-prepared spectrum waveform. It is now possible to roughly determine the cause of an anomaly by expressing multiple causes of anomaly using probabilities. Although this device may be effective as a stationary device for conducting academic research, it is not only incapable of making sufficient judgments necessary for on-site abnormality diagnosis, but also cannot be used with high accuracy and wide range. Because it performs analysis, it is large and difficult to carry around.

Therefore, in view of the above-mentioned conventional problems, it is an object of the present invention to provide a mechanical equipment abnormality diagnosing device with a simple configuration that can automatically diagnose abnormalities in mechanical equipment including their causes without manual intervention. It is an object.

[Means for solving problems]

In order to achieve the above object, the mechanical equipment abnormality diagnosis device according to the present invention is as shown in FIG.
A frequency analysis means 1 which obtains a power spectrum by frequency analyzing vibration data picked up from mechanical equipment, and a predetermined abnormality at each of a plurality of predetermined frequencies of interest from the power spectrum obtained by the frequency analysis means 1. measurement pattern forming means 2 for detecting a frequency exceeding the abnormal level and forming a measurement pattern represented by a frequency of interest exceeding the abnormal level based on the detection result; a first memory means 3 which is formed and stores in advance a plurality of abnormality patterns, each represented by a plurality of specific focused frequencies; a second mask pattern storing in advance a plurality of mask patterns for masking frequencies of interest other than the plurality of specific frequencies of interest corresponding to the abnormal item;
memory means 4 for storing the measurement pattern in the second memory means 4;
masking means 5 for sequentially masking each abnormal item with a mask pattern from the memory means 4 of
and a comparison means 6 for sequentially comparing the measurement pattern masked by the masking means 5 and the abnormality pattern for each abnormality item from the first memory means 3, The present invention is characterized in that the presence or absence of an abnormality for each abnormality item is determined based on the results.

[Effect]

In the above configuration, the vibration measurement results of the mechanical equipment are patterned based on whether the level of a predetermined frequency of interest is equal to or higher than a predetermined abnormality level, and is irrelevant to the determination of a specific abnormality in this measurement pattern. The measurement pattern portion corresponding to the frequency of interest is masked using a mask pattern prepared in advance for each abnormality item, and this masked measurement pattern is compared with the abnormality pattern prepared in advance for each abnormality item. Therefore, even if a complex abnormality including multiple abnormalities occurs or an unrelated frequency of interest exists, the presence or absence of an abnormality and the item of the abnormality can be automatically determined.

〔Example〕

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

FIG. 2 is a diagram showing the configuration of a system using the abnormality diagnosis device 10 according to the present invention, in which a vibration signal corresponding to vibrations generated by the vibration pickup 11 is applied to a vibration meter 12, where the vibration signal is The signal is converted into a signal representing the amplitude of the acceleration (or displacement, velocity) and is input to the device 10, and its magnitude is indicated by the deflection of the pointer 12a. device 10
A pen recorder 13, a sequencer 14 for controlling the mechanical equipment being diagnosed, for example, and a personal computer 15 are connected to.

The device 10 is made in the form of an attache case for easy portability, and FIG. 3 shows the front panel surface when the case lid is opened. In the figure, 10a is a power switch;
10b is an input signal range selection switch, 10c is an analog output waveform selection switch, 10d is a low-pass filter contact/disconnection switch, 10e is an envelope filter contact/disconnection switch, 10f is a battery check switch, 10g is a digital output transmission speed adjustment knob, 10h is AC100V power input terminal, 10i and 10j are analog signal input and output terminals, 10k
is a digital signal input/output terminal, 10l is a serial data output terminal, and analog signal input terminal 10
i has a vibration meter 12 and an analog signal output terminal 10.
A pen recorder 13 is connected to j, and a personal computer 15 is connected to the serial data output terminal 10l. 10m is a dot matrix liquid crystal display (LCD) that displays 32 characters x 4 lines.
10n is a key unit (KU) having a function key and a numeric keypad;
n is provided with indicators numbered 1 to 8 to instruct calculation in progress, abnormality, CPU abnormality, low pass, envelope, input ±2.5V, output 0 to 5V, and battery replacement.

Inside the case on the back side of the front panel above, there is a fourth
It has a built-in electrical circuit as shown in the block diagram in the figure. In the figure, 100 is a central processor unit (CPU) that performs various tasks according to predetermined programs, 102 is a read-only memory (ROM) that stores programs that cause the CPU 102 to perform predetermined tasks, and A memory device (MU) 104 consisting of a random access memory (RAM) that temporarily stores input data and data generated during a work process or as a result.
106 is a functional arithmetic unit (APU) dedicated to specific operations; 106 is a key interface;
8 is a digital input/output circuit (DIO), 110 is an analog-to-digital converter (ADC), 112 is a digital-to-analog converter (DAC), 114 is a parallel-serial data conversion circuit (PSC), 116 is an isolator, 118 is a low-pass filter, envelope A filter unit 120 consisting of a line filter is provided between the sections 100 to 114, and is a bus for transmitting and receiving address signals, data signals, and control signals between these sections.

With the above configuration, the apparatus 10 diagnoses abnormalities in mechanical equipment according to the programmed processing procedure as shown in the general flowchart of FIG.

When the program starts, first the vibration meter 12
Vibration is measured based on, for example, an acceleration signal having a waveform as shown in FIG. 6a, which is input to the analog signal input terminal 10i. For example, the acceleration signal input to the analog signal input terminal 10i is 3K.
It is sampled at a sampling frequency of Hz, and the sampled sample value is quantized into 12-bit data in the ADC 110. The resulting vibration data is sequentially stored in the RAM of the memory device 102.

Next, frequency analysis is performed on the stored vibration data. This frequency analysis uses the principle that a time-varying signal is made up of a combination of multiple regular repeating signals, and calculates the fundamental frequency and contribution size of those repeating signals, that is, the power spectrum. This allows fast processing of the Fourier transform F(w)=∫ ^∞ _-∞ f(t)e ^-jwt dt =∫ ^∞ _-∞ f(t)(cos wt−j sin wt)dt By performing FFT, a power spectrum as shown in FIG. 6b is obtained.

In the next step, frequencies with peaks are automatically detected from the FFT results. A method for this purpose may be to obtain a differential value and set the peak frequency at a point where the differential value is equal to or greater than a predetermined value and the sign of the differential value is reversed from 10 to 1.

In the subsequent step, a peak frequency that matches the frequency of interest, which is defined as a frequency at which the abnormal pattern found in the test can be determined, is detected. To detect the peak frequency that matches this frequency of interest, the peak frequency fn is the frequency of interest f ₀ ± detection width Δf ₀
This is done depending on whether or not it is within the 6th
Figure c is marked with o.

In the next step, abnormality is determined by determining the peak value. This step is to determine whether the peak value exceeds an abnormal level, and the abnormal level is determined to be n times the normal level (n is determined experimentally). The abnormal level at the frequency of interest shown in FIG. 6c and the frequency of interest exceeding the abnormal level are marked with ◎ in FIG. 6d, and this is the measurement pattern.

Next, abnormality items are determined by pattern determination based on the measurement pattern as the abnormality determination result. This abnormal term is determined by determining which pattern the combination of abnormal frequencies matches with which pattern is shown in an abnormal pattern table that has been experimentally determined and prepared in advance.

Assuming that the above-mentioned abnormality diagnosis device 10 is for diagnosing abnormalities in a rotating mechanism, the RAM of the memory device 102 stores installation abnormalities and bearings shown in FIGS. 7 and 8, respectively, as the abnormality pattern table. Information regarding abnormalities will be provided.

In the case of a rotating mechanism, 21 frequencies of interest are set as shown in the figure, f ₀ is the natural frequency due to the rotation speed of the rotating mechanism, Z is the number of rolling elements of the rotating mechanism, fc, fb,
fi is expressed by the following formula.

fc=1/2f ₀ (1-D1/D2cosα) fb=1/2D2/D1 {1-(D2/D1cosα) ² } fi=1/2f ₀ (1+D1/D2cosα) In the above equation, D1 is the rolling element The diameter, D2 is the bearing pitch circle diameter, α is the contact angle, f ₀ , Z, D1, D2,
α is F ₀ , Z, D1, D2 in key unit 10n,
A specific frequency is determined by inputting it in advance using the α key.

The presence or absence of the above-mentioned 21 frequencies of interest can be indicated by 0 or 1 of each of the 21 bits, but since the data is processed in units of 1 byte (8 bits) inside the microcomputer, processing is simplified. For convenience, we use 8 bits x 3 = 24 bits of information. Then, by dividing this into 4 bits each starting from the 1/3f ₀ side, each bit is divided using 24 bits.
It is expressed in hexadecimal. For example, for bearing eccentricity, it is [00, 01, 20] ₁₆ .

In order to determine the above-mentioned abnormality level, it is necessary to know the level at the frequency of interest when the mechanical equipment is normal. FIG. 9 shows a programmatic flowchart for determining this normal level. When the program starts, it first measures the amplitude of acceleration and imports the 1024 sampled vibration data obtained from the measurement. Next, frequency analysis is performed on the captured data. At the end of this frequency analysis, it is determined whether or not it has finished six times, and this determination is
Repeat the above operations until you get YES. Judgment
If YES, then the results of the six frequency analyzes are added and the average is calculated. After that, the average level of the measurement frequency matching the frequency of interest is captured. This is 1
It checks whether one measurement frequency matches the 21 frequencies of interest, and if they match, the level is taken in. This is done for 512 measurement frequencies.

By examining how many times the normal level obtained above is during an artificially caused abnormality for a specific frequency of interest, we can calculate the constant n by which the normal level is multiplied to obtain the abnormal level. is determined.

Next, FIG. 10 shows a flowchart of a program for diagnosing an abnormality. In this flowchart, when the program starts, the amplitude of the vibration acceleration signal obtained based on the vibration signal picked up from the mechanical equipment to be diagnosed is first measured and 1024 pieces of vibration data are captured. Next, we perform frequency analysis on the captured vibration data,
After performing this six times, the added values are averaged. The above is the same as the flowchart shown in FIG. 9 for determining the normal level.

Next, a frequency having a peak value is detected, and then one of the detected peak frequencies that matches the frequency of interest is detected. Subsequently, an abnormality is determined by determining whether the level of the peak frequency that matches the frequency of interest exceeds an abnormal level. Based on the frequency of the abnormal level obtained in this way, next, the abnormal item is determined by pattern judgment. Then, the abnormal item is notified by display or the like, and the execution of the program is terminated.

In order to determine the above-mentioned abnormality items, it is necessary to determine which of the abnormality pattern tables prepared in advance matches the data regarding the presence or absence of abnormalities for the 21 frequencies of interest obtained through the above-mentioned abnormality determination. Just check to see if there are any. However, the above data may include information about a plurality of abnormal items, and in such cases, it is not possible to find the abnormal item by comparing the corresponding bits of the data and the abnormal pattern.

FIG. 11 shows a flowchart of a program for determining the above-mentioned abnormality items. In this flowchart, first, data obtained by abnormality determination is masked with a mask pattern selected from a mask pattern table prepared in advance corresponding to the abnormality item. Now, if the measurement pattern is [00, 00, 7f] ₁₆ ,
This measurement pattern is first masked with a mask pattern [00, 00, 3F] ₁₆ for abnormal items called "unbalance." This masking process temporarily erases unnecessary information from the measured frequency pattern for comparison with the abnormal pattern at hand by taking the AND of each bit of the mask bit pattern and the measured frequency pattern. It is something to do. In this example, the measurement pattern after this mask is [00, 00,
3F〕 ₁₆ .

Thereafter, a comparison is made to see whether the measurement pattern on the mask side matches the abnormal pattern of the corresponding abnormal item. The abnormal pattern of the above abnormal item "Unbalance" is [00, 00, 34] ₁₆ , [00, 00,
There are four, 14] ₁₆ , [00, 00, 24] ₁₆ , and [00, 00, 04] ₁₆ , each of which is contrasted with [00, 00, 3F] ₁₆ .

Subsequently, a determination is made as to the result of the comparison. In the above example, the answer is NO, and in this case, it is determined whether this is the final abnormality item to be determined. In the above example it is NO,
In this case, the process advances to the next abnormal item, which is masked in the same manner as described above, and then compared with the abnormal pattern. When the above processing is performed in sequence and the above measurement pattern is multiplied by a mask pattern [00, 01, 30] ₁₆ for the abnormality item "bearing eccentricity", a pattern [00, 00, 30] ₁₆ is obtained after masking. When this is compared with the bearing eccentricity abnormality pattern, some matches are found, and the matched abnormality items are later stored in the RAM. Finally, since the measurement pattern has a peak in addition to this bearing eccentricity, it is also memorized that there are other abnormalities.

After storing the above-mentioned abnormal item, it is determined whether it is the final item to be judged, and if YES, the next stored abnormal item is displayed and the whole process ends. Of course, the determination results of the above-mentioned abnormality items can be kept as a record if necessary.

FIG. 12 is a diagram summarizing the above, and what is shown in the blocks is the data written to the RAM.

First, data within a block is prepared, and the frequency of interest shown in the block is calculated using specific data within the block. Under these conditions, amplitude measurements are performed at a sampling frequency of 3KHz, and the results are analog-to-digital converted and collected as 1024 pieces of vibration data. The collected vibration data is frequency-analyzed to obtain a power spectrum of 512 frequencies, the above process is repeated six times, and the average is obtained.

A peak frequency is detected from this power spectrum, and a frequency within a predetermined range of the frequency of interest is obtained from the detected peak frequency. Furthermore, a pattern exceeding the abnormal level is obtained and used as a measurement pattern.
Thereafter, a masked measurement pattern is obtained, and this is compared with the abnormal pattern to determine the presence or absence of abnormal items. This mask and abnormal item determination are sequentially repeated for all items, and the results are output as an abnormal message.

〔effect〕

As explained above, according to the present invention, it is only necessary to perform data processing on a predetermined frequency of interest, and it is possible to simply compare the measurement pattern after masking with the abnormal pattern, even if a complex abnormality occurs. Since the items of abnormality can be determined automatically and extremely easily, the amount of data to be processed can be significantly reduced, and the device configuration can also be simplified.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the basic configuration of the device according to the present invention, FIG. 2 is a configuration diagram of a mechanical equipment abnormality diagnosis system using the device according to an embodiment of the present invention, and FIG. 3 is a block diagram showing the basic configuration of the device according to the present invention. FIG. 4 is a block diagram showing the electric circuit configuration in the device shown in FIG. 1, FIG. 5 is a general flowchart showing the work performed by the CPU in FIG. 4, and FIG.
The figure shows the data of the processing process by the CPU, Figures 7 and 8 show abnormal patterns for different abnormal states of the rotating mechanism, and Figure 9 is a flowchart for collecting normal level data. FIG. 10 is a flowchart for obtaining measurement patterns, FIG. 11 is a flowchart for determining abnormal items, and FIG. 12 summarizes the data stored in the memory device in FIG. FIG. DESCRIPTION OF SYMBOLS 1... Frequency analysis means, 2... Measurement pattern forming means, 3... First memory means, 4... Second memory means, 5... Mask means, 6... Comparison means.

Claims (1)

[Scope of Claims] 1. Frequency analysis means for obtaining a power spectrum by frequency-analyzing vibration data picked up from mechanical equipment, and each of a plurality of frequencies of interest predetermined from the power spectrum obtained by the frequency analysis means. a measurement pattern forming means for detecting an abnormality exceeding a predetermined abnormality level in the equipment, and forming a measurement pattern represented by a frequency of interest exceeding the abnormality level based on the detection result; first memory means formed in correspondence with each other and storing in advance a plurality of abnormality patterns, each represented by a plurality of specific frequencies of interest; and a first memory means formed in correspondence with the plurality of abnormality items, respectively; a second memory means storing in advance a plurality of mask patterns for masking frequencies of interest other than the plurality of specific frequencies of interest, each of which corresponds to the abnormality item; a masking means for sequentially masking each abnormality item with a mask pattern; and sequentially comparing the measurement pattern masked by the masking means with the abnormality pattern for each abnormality item from the first memory means. A mechanical equipment abnormality diagnosing device comprising: a comparing means, and determining whether or not there is an abnormality in each of the abnormality items based on the comparison result of the comparing means.