Anomaly detection device and method

Abstract

To detect an abnormal state of a motor with higher accuracy compared with the conventional art.SOLUTION: A device for detecting an abnormality comprises a signal processing unit. The signal processing unit performs frequency analysis of data of a current or a voltage in first and second time periods different from each other to retrieve a first drive frequency corresponding to a maximum peak value of a spectrum in the first time period and a second drive frequency corresponding to a maximum peak value of a spectrum in the second time period. The signal processing unit corrects a frequency axis of one spectrum so that the respective maximum peak values of the respective spectra are coincident with each other on the basis of a first spectrum having the first drive frequency and a spectrum having the second drive frequency, and then, calculates an average spectrum related to the corrected spectrum and the uncorrected spectrum. The signal processing unit retrieves an abnormal peak value equal to or more than a predetermined threshold at a frequency different from the drive frequency having the maximum peak value on the basis of the average spectrum, to determine an abnormal state of a motor depending on presence or absence of the abnormal peak value.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an abnormality detection device and method for detecting an abnormal state of a rotating machine such as a motor. [Background technology]

[0002] In monitoring equipment that diagnoses the condition of a motor, when data obtained by a sensor, for example, is frequency-analyzed, if noise components are found in addition to the peaks that always appear, this is determined to be an abnormal condition, and the noise components are quantified to indicate the degree of abnormality.

[0003] For example, Patent Document 1 discloses an abnormality detection device capable of identifying a frequency corresponding to the period of a sound emitted by rotation based on an audio signal even in the presence of periodic noise. This abnormality detection device detects the envelope of an audio signal representing a periodic sound emitted from a rotating body having a predetermined number of blades and a periodic sound emitted from other objects, calculates a frequency spectrum of the audio signal from the envelope for each frame, and detects a candidate frequency corresponding to the period of the sound emitted from the rotating body in each frame. Next, for each frame, the device finds a duration during which the power fluctuation of the power of the frequency spectrum components in the candidate detected for that frame is equal to or less than a certain value, and identifies the candidate with the longest duration as the frequency corresponding to the period of the sound emitted from the rotating body. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-100975 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, the above-mentioned conventional technology is based on the premise that the rotation speed of the motor is constant, and if the drive frequency of the motor fluctuates during the measurement period, it becomes difficult to distinguish between the peaks that always appear and the noise components. As a result, there is a problem that the peaks that always appear are judged as noise components, and the degree of abnormality in the diagnosis result becomes high.

[0006] In order to obtain accurate diagnostic results, the user must keep the motor rotation speed constant at least while performing diagnostic measurements. Also, if the diagnostic results indicate a large degree of abnormality, the user must determine whether this is due to an actual motor abnormality or a change in the motor rotation speed.

[0007] SUMMARY OF THE PRESENT EMBODIMENT An object of the present invention is to solve the above problems and provide an abnormality detection device and method capable of detecting an abnormal state of a motor with higher accuracy than the prior art. [Means for solving the problem]

[0008] An abnormality detection device according to one aspect of the present invention comprises: 1. An abnormality detection device that detects an abnormal state of a motor based on a current or voltage supplied to the motor from a power source, comprising: performing a frequency analysis of the current or voltage data for different first and second time periods to search for a first drive frequency corresponding to a maximum peak value of the spectrum for the first time period and a second drive frequency corresponding to a maximum peak value of the spectrum for the second time period; A first spectrum having the first driving frequency and a second spectrum having the second driving frequency. Second correcting a frequency axis of one spectrum based on the first spectrum so as to match the maximum peak values ​​of the first spectrum and then calculating an average spectrum relating to the corrected spectrum and the uncorrected spectrum; a signal processing unit that searches for an abnormal peak value equal to or greater than a predetermined threshold value at a frequency different from a drive frequency having a maximum peak value based on the averaged spectrum, and judges an abnormal state of the motor depending on the presence or absence of the abnormal peak value; Prepare. Effect of the Invention

[0009] Therefore, the abnormality detection device according to the present invention can detect an abnormal state of a motor with higher accuracy than the prior art. [Brief description of the drawings]

[0010] [Figure 1] 1 is a block diagram showing an example of the configuration of a motor abnormality detection device 4 according to a first embodiment. [Diagram 2] 4 is a flowchart showing an abnormality detection process executed by the processor 10 of FIG. [Diagram 3] 3 is a graph showing an example of a plurality of spectra of the motor current calculated when the processor 10 of FIG. 1 executes the abnormality detection process of FIG. 2. [Figure 4A] 10 is a flowchart showing abnormality detection pre-processing executed by a processor 10 of a motor abnormality detection device 4 according to a second embodiment. [Figure 4B] 10 is a flowchart showing an abnormality detection process executed by a processor 10 of a motor abnormality detection device 4 according to a second embodiment. [Figure 5A] 4C is a graph showing a first example of a spectrum of a motor current calculated when the processor 10 of FIG. 1 executes the abnormality detection process of FIG. 4B. [Figure 5B] 4C is a graph showing a second example of the spectrum of the motor current calculated when the processor 10 of FIG. 1 executes the anomaly detection process of FIG. 4B. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Note that the same or similar components are denoted by the same reference numerals.

[0012] (Embodiment 1) Fig. 1 is a block diagram showing an example of the configuration of a motor abnormality detection device 4 according to embodiment 1. Fig. 1 shows the abnormality detection device 4 and its peripheral circuits, in which AC power from an AC power source 1 is supplied to a motor 2 via a current sensor 3. In Fig. 1, the abnormality detection device 4 is configured to include an AD converter 11, a processor 10 having a frequency analysis unit 12 and an abnormality detection unit 13 to configure a "signal processing unit", and a display unit 14. Here, the frequency analysis unit 12 and the abnormality detection unit 13 have memories 12m and 13m, respectively.

[0013] In FIG. 1, a current sensor 3 detects the current value of a current supplied from an AC power source 1 to a motor 2, and outputs a detection signal indicating the current value to an AD converter 11. The AD converter 11 converts the input detection signal into current data indicating the current value, and then outputs the current data to a frequency analysis unit 12. The frequency analysis unit 12 performs a fast Fourier transform (FFT) process on the input current data over a predetermined time period T1 and two divided time periods T2 and T3. The abnormality detection unit 13 judges an abnormal state of the motor based on the spectrum after the FFT process, and displays the judgment result on a display unit 14. Specifically, the processor 10 consisting of the frequency analysis unit 12 and the abnormality detection unit 13 judges an abnormal state of the motor by executing the abnormality detection process of FIG. 2, and displays the judgment result on a display unit 14.

[0014] Fig. 2 is a flowchart showing an abnormality detection process executed by the processor 10 of Fig. 1. Also, Fig. 3 is a graph showing an example of a plurality of spectra of the motor current calculated when the processor 10 of Fig. 1 executes the abnormality detection process of Fig. 2.

[0015] The abnormality detection process according to the first embodiment will be described below with reference to FIGS.

[0016] In step S1 of FIG. 2, current data of the AD converted value for a predetermined time period T1 detected by the current sensor 3 is input, and in step S2A, the frequency analysis unit 12 time The AD converted value of the current value during the period T1 is subjected to FFT processing using the memory 12m to calculate the reference spectrum 100 (FIG. 3(a)) of the FFT value (power). time The AD-converted value of the current value during period T1 is divided into two predetermined time periods T2 and T3, and for each of the time periods T2 and T3, FFT processing is performed using memory 12m to calculate the spectra 101 and 102 (Figures 3(b) and 3(c)) of the FFT values ​​(power).

[0017] Next, in step S3, the abnormality detection unit 13 searches for a drive frequency corresponding to the maximum peak value of each of the time periods T2 and T3 based on the spectrum of each of the time periods T2 and T3. Note that the processes of steps S3 to S9 are executed by the abnormality detection unit 13. In step S4, the lower drive frequency f low (In Fig. 3, f low =51Hz) at high drive frequency f high (In FIG. 3, f high = 52Hz) low / f high ) at a high driving frequency f high The frequency axis of the spectrum 102 having a high driving frequency f high By correcting the frequency axis of the spectrum 102 having the corrected high driving frequency f high Then, in step S5, the spectrum 102A (FIG. 3(d)) having the corrected high driving frequency f high and a spectrum 102A having a low drive frequency f low The spectrum 101 having the maximum peak value is As if By adding these two spectra and dividing them by two, an average spectrum 103 (FIG. 3(e)) is calculated and stored in memory 13m.

[0018] Next, in step S6, the abnormality detection unit 13 searches for a peak value of an abnormal peak 104 equal to or greater than a predetermined threshold value at a frequency different from the driving frequency having the maximum peak value based on the averaged spectrum 103 (FIG. 3(e)). Here, the abnormal peak 104 has a peak value having, for example, the second or third strongest power after the maximum peak. Furthermore, in step S7, the abnormality detection unit 13 judges whether or not there is an abnormal peak, and if YES, proceeds to step S8, whereas if NO, proceeds to step S9. In step S8, it is judged that the motor is in an abnormal state, and the judgment result is displayed on the display unit 14, and the abnormality detection process is terminated. Also, in step S9, it is judged that the motor is not in an abnormal state, and the judgment result is displayed on the display unit 14, and the abnormality detection process is terminated.

[0019] In the abnormality detection process in Fig. 2, the motor drive current data acquired during a given time period T1 is divided into two time periods T2 and T3 for frequency analysis, and the drive frequency that is the maximum peak value for each of the two divided time periods T2 and T3 is searched for. Of the two drive frequencies found, the lower drive frequency divided by the higher drive frequency is multiplied by the frequency axis of the spectrum of the frequency analysis result having the higher drive frequency, and this is added to the spectrum of the frequency analysis result having the lower drive frequency and averaged.

[0020] For example, in the embodiment of Fig. 3, if the driving frequency fluctuates over 4 seconds of data, causing the peaks of abnormality characteristics to become dull, for example, if an FFT is performed between 0 and 2 seconds and a frequency analysis is performed between 2 and 4 seconds, and the maximum peak frequencies are 51 Hz and 52 Hz, respectively, the spectrum obtained by multiplying the frequency axis of 52 Hz by 51 / 52 is added to the spectrum of the frequency analysis result of 51 Hz and averaged. This makes the abnormality characteristics appear sharper, and allows for more sensitive and accurate capture of abnormality characteristics than envelope FFT.

[0021] As described above, by performing the anomaly detection process in Fig. 2, it is possible to capture peak 104 (Fig. 3(e)), which was not visible in the conventional technology, as shown in Fig. 3(e). In particular, when the drive frequency fluctuates, it becomes easy to distinguish between peaks that constantly appear and noise components, and the anomaly detection process can be performed with higher accuracy than in the conventional technology.

[0022] (Embodiment 2) Fig. 4A is a flowchart showing abnormality detection pre-processing executed by the processor 10 of the motor abnormality detection device 4 according to embodiment 2. Fig. 4B is a flowchart showing abnormality detection processing executed by the processor 10 of the motor abnormality detection device 4 according to embodiment 2. The abnormality detection processing according to embodiment 2 differs from the abnormality detection processing according to embodiment 1 in the following points. The abnormality detection device 4 is configured similarly to that shown in Fig. 1. (1) The anomaly detection pre-processing in FIG. 4A is executed at an earlier stage than the anomaly detection processing in FIG. 4B. As a result, the anomaly detection unit 13 calculates the peak width (frequency width) having the maximum peak value for an FFT value equal to or greater than a predetermined threshold value, and determines the low cut-off frequency fc based on the calculated peak width and stores it in the memory 13m. (2) The abnormality detection process in FIG. 4B is characterized in that the process of step S2C is inserted between steps S2A and S2B, as compared with the abnormality detection process in FIG. The differences will be explained below.

[0023] In step S11 of the abnormality detection pre-processing in FIG. 4A, an AD converted value of a current value during a predetermined time period T1 detected by the current sensor 3 is input. Next, in step S12, the frequency analysis unit 12 time The AD converted values ​​of the current values ​​during the period T1 are subjected to FFT processing using the memory 12m to calculate the spectrum of the FFT values ​​(power). Furthermore, in step SS13, the frequency width of the peak having the maximum peak value is calculated for FFT values ​​equal to or greater than a predetermined threshold value, and in step S14, the low cutoff frequency fc is set based on the calculated frequency width of the peak, and the abnormality detection pre-processing is terminated.

[0024] In step S14, the detection sensitivity is increased by setting the lower cutoff frequency fc so that the lower cutoff frequency fc is increased as the calculated peak width (frequency width) increases.

[0025] In step S2C of the anomaly detection process in FIG. 4B, a filtering process is performed on the reference spectrum 100 using a high-pass filter having a lower cutoff frequency fc determined in the anomaly detection pre-processing, and the spectrum after the filtering process is used as the reference spectrum 100, and steps S2B to S9 are executed in the same manner as in FIG. 2.

[0026] Fig. 5A is a graph showing a first example of the spectrum of the motor current calculated when the abnormality detection process of Fig. 4B is executed by the processor 10 of Fig. 1. Also, Fig. 5B is a graph showing a second example of the spectrum of the motor current calculated when the abnormality detection process of Fig. 4B is executed by the processor 10 of Fig. 1.

[0027] In this second embodiment, when the current data of the motor drive current is frequency-analyzed to detect the characteristics of the motor abnormality, unnecessary DC components and low-frequency components in the vicinity are cut off, i.e., high-pass filtering is performed. Here, when the motor drive frequency fluctuates, the frequency width of the DC component and its vicinity may become large, and some of them cannot be cut off and are detected as the characteristics of the abnormality. Therefore, the bandwidth for cutting low-frequency components is adjusted according to the degree of fluctuation in the drive frequency to increase the detection sensitivity. For example, the spectrum of FIG. 5A is R As shown in spectrum 201, when the variation in the drive frequency is within ±0.5 Hz, the lower cutoff frequency fc is set to 1 Hz. Also, as shown in spectrum 202 in Fig. 5B, when the variation in the drive frequency is within ±1.0 Hz, it is preferable to set the lower cutoff frequency fc to 2 Hz.

[0028] As described above, according to the second embodiment, in the anomaly detection process of the first embodiment, the calculated spectrum is subjected to high-pass filtering using a predetermined low-frequency cutoff frequency fc, thereby removing unnecessary N The DC component and low frequency component are cut off. Here, the detection sensitivity can be increased by setting the low cutoff frequency fc so that the lower cutoff frequency fc is increased as the frequency width of the calculated peak increases. This makes it easier to distinguish between peaks that constantly appear and noise components, particularly when the drive frequency fluctuates, and allows the abnormality detection process to be performed with higher accuracy than in the prior art.

[0029] (Modification) In the above embodiment, the current data of the current supplied from the AC power supply to the motor is subjected to frequency analysis to determine whether the motor is in an abnormal state. However, the present invention is not limited to this. The voltage data of the voltage supplied from the AC power supply to the motor may be subjected to frequency analysis to determine whether the motor is in an abnormal state in a similar manner.

[0030] In the above embodiment, the frequency axis of the spectrum having the high drive frequency is corrected so that the frequency axis of the spectrum having the high drive frequency is multiplied by the division value obtained by dividing the lower drive frequency by the higher drive frequency of two drive frequencies including the first and second drive frequencies, thereby calculating a spectrum having the high drive frequency after the correction, and adding up the maximum peak values ​​of the two spectra based on the spectrum having the high drive frequency after the correction and the spectrum having the low drive frequency. As if However, the present invention is not limited to this, and the frequency axis of the spectrum having the low drive frequency is corrected so that the frequency axis of the spectrum having the low drive frequency is multiplied by the division value obtained by dividing the higher drive frequency by the lower drive frequency of the first and second drive frequencies, thereby calculating a spectrum having the low drive frequency after the correction, and adding up the maximum peak values ​​of the two spectra based on the spectrum having the low drive frequency after the correction and the spectrum having the high drive frequency. As ifThe average spectrum may be calculated.

[0031] Therefore, in the present invention, a first spectrum having the first driving frequency and a second spectrum having the second driving frequency are Second Based on the spectra, the frequency axis of one of the spectra may be corrected so as to match the maximum peak values ​​of the spectra, and then an average spectrum of the corrected spectrum and the uncorrected spectrum may be calculated. [Industrial Applicability]

[0032] As described above in detail, according to the present invention, when the drive frequency fluctuates, it becomes easy to distinguish between peaks that constantly appear and noise components, and abnormality detection processing can be performed with higher accuracy than with conventional techniques. [Explanation of symbols]

[0033] 1 AC power supply 2 Motors 3 Current Sensor 4. Anomaly detection device 10 Processor (signal processing section) 11 AD converter 12 Frequency Analysis Section 12M memory 13 Anomaly detection section 13M memory 14 Display section

Claims

1. 1. An abnormality detection device that detects an abnormal state of a motor based on a current or voltage supplied to the motor from a power source, comprising: performing frequency analysis on the current or voltage data for different first and second time periods to search for a first drive frequency corresponding to a maximum peak value of the spectrum for the first time period and a second drive frequency corresponding to a maximum peak value of the spectrum for the second time period; based on a first spectrum having the first drive frequency and a second spectrum having the second drive frequency, correcting a frequency axis of one spectrum so as to match maximum peak values ​​of the spectra, and then calculating an average spectrum relating to the corrected spectrum and the uncorrected spectrum; a signal processing unit that searches for an abnormal peak value equal to or greater than a predetermined threshold value at a frequency different from a drive frequency having a maximum peak value based on the averaged spectrum, and judges an abnormal state of the motor depending on the presence or absence of the abnormal peak value; Equipped with an abnormality detection device.

2. When calculating the average spectrum, the signal processing unit (1) calculating a spectrum having a higher driving frequency after correction by correcting a frequency axis of the spectrum having the higher driving frequency so as to multiply the frequency axis of the spectrum having the higher driving frequency by a value obtained by dividing a lower driving frequency by a higher driving frequency of the first and second driving frequencies, and calculating the average spectrum based on the spectrum having the higher driving frequency after correction and the spectrum having the lower driving frequency so as to equalize the maximum peak values ​​of the two spectra; or (2) calculating a spectrum having a lower drive frequency after correction by correcting a frequency axis of the spectrum having the lower drive frequency so as to multiply the frequency axis of the spectrum having the lower drive frequency by a value obtained by dividing a higher drive frequency by a lower drive frequency of the first and second drive frequencies, and calculating the average spectrum based on the spectrum having the lower drive frequency after correction and the spectrum having the higher drive frequency so as to match the maximum peak values ​​of the two spectra; The abnormality detection device according to claim 1 .

3. the signal processing unit performs high-pass filtering on the current or voltage data to be frequency-analyzed at a predetermined lower cutoff frequency before performing frequency analysis over the first and second time periods; The abnormality detection device according to claim 1 or 2.

4. The signal processing unit includes: performing a frequency analysis on the current or voltage data for a time period including the first and second time periods, calculating a frequency width of a peak having a maximum peak value based on a spectrum of the frequency analysis result, and setting the lower cutoff frequency based on the calculated frequency width of the peak. The abnormality detection device according to claim 3 .

5. the signal processing unit sets the lower cutoff frequency to be higher as the frequency width of the calculated peak becomes wider. The abnormality detection device according to claim 4.

6. 1. An abnormality detection method for detecting an abnormal state of a motor based on a current or voltage supplied to the motor from a power supply, comprising: A signal processing unit performs frequency analysis on the current or voltage data for different first and second time periods to search for a first driving frequency corresponding to a maximum peak value of the spectrum in the first time period and a second driving frequency corresponding to a maximum peak value of the spectrum in the second time period; a step in which the signal processing unit corrects a frequency axis of one spectrum based on a first spectrum having the first drive frequency and a second spectrum having the second drive frequency so as to match maximum peak values ​​of the spectra, and then calculates an average spectrum relating to the corrected spectrum and the uncorrected spectrum; the signal processing unit searches for an abnormal peak value equal to or greater than a predetermined threshold value at a frequency different from a drive frequency having a maximum peak value based on the averaged spectrum, and judges an abnormal state of the motor depending on the presence or absence of the abnormal peak value; Anomaly detection methods.

7. The step of calculating the average spectrum comprises the steps of: (1) calculating a spectrum having a higher driving frequency after correction by correcting a frequency axis of the spectrum having the higher driving frequency so as to multiply the frequency axis of the spectrum having the higher driving frequency by a value obtained by dividing a lower driving frequency by a higher driving frequency of the first and second driving frequencies, and calculating the average spectrum based on the spectrum having the higher driving frequency after correction and the spectrum having the lower driving frequency so as to equalize the maximum peak values ​​of the two spectra; or (2) calculating a spectrum having a lower drive frequency after correction by correcting a frequency axis of the spectrum having the lower drive frequency so as to multiply the frequency axis of the spectrum having the lower drive frequency by a value obtained by dividing a higher drive frequency by a lower drive frequency of the first and second drive frequencies, and calculating the average spectrum based on the spectrum having the lower drive frequency after correction and the spectrum having the higher drive frequency so as to match the maximum peak values ​​of the two spectra; The anomaly detection method according to claim 6.

8. a step of high-pass filtering the current or voltage data to be frequency-analyzed at a predetermined lower cutoff frequency by the signal processing unit before the frequency analysis is performed over the first and second time periods; The anomaly detection method according to claim 6 or 7, further comprising:

9. a step of the signal processing unit frequency-analyzing the current or voltage data for a time period including the first and second time periods, calculating a frequency width of a peak having a maximum peak value based on a spectrum of the frequency analysis result, and setting the lower cutoff frequency based on the calculated frequency width of the peak; The anomaly detection method of claim 8 further comprising:

10. the step of setting the lower cutoff frequency includes setting the lower cutoff frequency to a higher value as the frequency width of the calculated peak becomes wider; The anomaly detection method according to claim 9.

Images