JP7717664B2 - Device and method for diagnosing abnormalities in mechanical equipment - Google Patents

Description

The present invention relates to a device and method for diagnosing abnormalities in rotating machinery that rotates at low speeds, such as sludge collectors used in sewage treatment plants.

As shown in Figure 9, the sludge collector 41 is composed of a drive unit 43, an endless chain 45, a sprocket 47, and flights 49. The flights 49 are attached at regular intervals to the endless chain 45, which is driven via a drive shaft and a driven shaft sprocket, moving the flights 49 in contact with the rail surface of the sedimentation tank 51, continuously scraping the sludge that has settled in the sedimentation tank 51 into a sludge hopper.

The sludge collector 41 shown in Figure 9 has sprockets 47 at five locations: a drive sprocket 53, a drive shaft drive sprocket 54 (coaxial with the driven shaft driven sprocket 55 but with a slightly larger outer diameter), and drive shaft driven sprockets 55, 57, 59, and 61.

Such abnormal phenomena in the sludge collector 41 include elongation of the endless chain 45 and wear of the sprocket 47 and flights 49.

However, the rotation speed of the drive unit 43 is very low, for example, about 0.4 rpm, and the flight 49 travels very slowly, for example, at 0.3 m/min, to scrape up the sludge. Furthermore, since most of the device is underwater, it is difficult to detect abnormalities using the five senses from outside the sedimentation basin 51.

Therefore, as a method for diagnosing abnormalities in such sludge collectors, for example, Patent Document 1 discloses a method in which a vibration sensor is attached to a drive unit installed outside the sedimentation tank and abnormalities are diagnosed based on the vibration waveform obtained from the vibration sensor.
This is based on the idea that the sprockets, flights, and endless chain are all connected by contact, etc., and any abnormal vibrations generated in any of the equipment are transmitted to the driving device on the ground via the endless chain.

The technology disclosed in Patent Document 1 is useful for diagnosing abnormalities in equipment when the equipment to be diagnosed is specified.

However, existing sludge collectors are used in many wastewater treatment facilities, and the scale and structure of the facilities (e.g., two-tier structure, three-tier structure, etc.) vary, and naturally the sizes and shapes of the sprockets, flights, and endless chains used therein also vary.
Therefore, even if an abnormality occurs in a sprocket, the frequency band and period of the vibration waveform detected by a vibration sensor attached to the drive device will be completely different between one piece of equipment and another piece of equipment.
For this reason, it is difficult to determine whether or not a failure has occurred in the equipment, and it is difficult for the technology disclosed in Patent Document 1 to address this issue.

In such cases, it is useful to capture the vibration waveform over time to determine whether or not there is an abnormality. For example, one possible approach is to use the wavelet transform disclosed in Patent Documents 2 and 3. Wavelet transform is a frequency analysis method that analyzes what frequencies are occurring over time and at what amplitudes.

JP 2018-153784 A JP 2010-190901 A Japanese Patent Application Laid-Open No. 2008-292288

However, the diagnostic method disclosed in Patent Document 2 involves an inspector examining the time-frequency distribution of wavelets to assess the remaining life of a bearing, which means that the assessment is dependent on the skill and knowledge of the inspector and may lack objectivity.

Furthermore, the diagnostic device disclosed in Patent Document 3 extracts amplitude values of multiple frequency bands using wavelet transform, but this is premised on processing with a focus on specific frequencies resulting from the known structure and rotation speed of the bearing.
Therefore, it is difficult to diagnose the presence or absence of abnormalities in equipment such as the sludge collector that is the subject of the present invention, which varies in size and structure and for which the frequency band of the target vibration waveform is unknown.

Note that the above-mentioned issues are common to equipment with unknown frequency bands, etc., and the equipment to be diagnosed is not limited to the sludge collector shown as an example.

The present invention was made to solve this problem, and aims to provide a device and method for diagnosing the presence or absence of abnormalities in mechanical equipment that can be applied even when the frequency band of the vibration waveform, etc., is unknown.

(1) The abnormality diagnosis device for mechanical equipment according to the present invention diagnoses the presence or absence of an abnormality in mechanical equipment,
The apparatus includes a vibration waveform acquisition unit that acquires vibration waveforms for a predetermined period of time using a vibration sensor installed in the target equipment, and an abnormality presence/absence diagnosis unit that analyzes the vibration waveform data acquired by the vibration waveform acquisition unit and diagnoses the presence or absence of an abnormality,
The abnormality diagnosis unit
wavelet transform means for performing wavelet transform on the collected vibration waveform to obtain time variations of wavelet coefficients for all frequencies;
a wavelet coefficient time waveform acquisition means for acquiring a time waveform of a wavelet coefficient at each frequency at a predetermined interval from the time change of the wavelet coefficient obtained by the wavelet transform;
an autocorrelation function calculation means for calculating an autocorrelation function for the time waveform of the wavelet coefficients of each frequency;
a correlation peak extraction means for extracting an arbitrary number of peak values of the correlation degree in the autocorrelation function for each frequency;
a determination means for determining whether or not abnormal periodic vibration has occurred based on the relationship between the peak value extracted for each frequency and time;
The present invention is characterized by the following features.

(2) Furthermore, in the device described in (1) above, the determination means creates a distribution diagram showing the relationship between the extracted peak values and time, calculates the center of gravity of the distribution diagram, compares it with a predetermined normal center of gravity, and determines whether or not there is an abnormality based on the degree of deviation between the two.

(3) A method for diagnosing the presence or absence of an abnormality in mechanical equipment according to the present invention is a method for diagnosing the presence or absence of an abnormality in mechanical equipment,
a vibration waveform acquisition step of acquiring a vibration waveform for a predetermined period of time using a vibration sensor installed in the target facility;
a wavelet transform process for performing wavelet transform on the collected vibration waveform to obtain time variations of wavelet coefficients for all frequencies;
a wavelet coefficient time waveform acquisition step of acquiring a time waveform of the wavelet coefficient at each frequency at a predetermined interval from the time change of the wavelet coefficient obtained by the wavelet transform;
an autocorrelation function calculation step of calculating an autocorrelation function for the time waveform of the wavelet coefficients of each frequency;
a correlation peak extraction step of extracting an arbitrary number of peak values of the correlation degree in the autocorrelation function for each of the frequencies;
a determination step of determining whether or not abnormal periodic vibration has occurred based on the relationship between the peak value extracted for each frequency and time;
The present invention is characterized by the following features.

(4) Furthermore, in the device described in (3) above, the determination step is characterized in that a distribution diagram showing the relationship between the extracted peak value and time is created, the center of gravity of the distribution diagram is calculated, and the presence or absence of an abnormality is determined based on the degree of deviation between the two, by comparing the center of gravity with a predetermined normal value.

In this invention, the time waveforms of wavelet coefficients are acquired for all frequencies, and the presence or absence of abnormal periodic vibrations is determined based on the time waveforms of the wavelet coefficients. This makes it possible to diagnose the presence or absence of abnormalities in mechanical equipment where the frequency band of the vibration waveform is unknown.

1 is an explanatory diagram of an abnormality presence/absence diagnosis device for mechanical equipment according to an embodiment of the present invention; 1 is a flowchart of a method for diagnosing the presence or absence of an abnormality in mechanical equipment according to an embodiment of the present invention. 2 is an example of an acceleration waveform obtained by the abnormality diagnosis device of FIG. 1. 4A and 4B are graphs showing an example of time changes in wavelet coefficients obtained by wavelet transforming a sampled acceleration waveform, where FIG. 4A shows an abnormal state and FIG. 4B shows a normal state. 10 is a graph showing time waveforms of wavelet coefficients for each frequency at a predetermined interval. 10 is a graph showing an autocorrelation function for the time waveform of wavelet coefficients of each frequency. FIG. 7 is an explanatory diagram (part 1) of a method for determining whether or not an abnormality exists based on the graph shown in FIG. 6 . FIG. 7 is an explanatory diagram (part 2) of a method for determining whether or not an abnormality exists based on the graph shown in FIG. 6 . FIG. 1 is an explanatory diagram illustrating the overall structure of a typical sludge scraper.

An abnormality diagnosis device for mechanical equipment according to one embodiment of the present invention (hereinafter simply referred to as an "abnormality diagnosis device") will be described using a sludge collector as an example of the mechanical equipment to be diagnosed.
First, the abnormality diagnosis device will be outlined with reference to Fig. 1, and then the abnormality diagnosis method will be explained. In Fig. 1, the same reference numerals as those in Fig. 9 are used to indicate the respective components of the sludge collector 41.

As shown in FIG. 1 , the mechanical equipment abnormality diagnosis device 1 of this embodiment includes a vibration waveform collection unit 3 and an abnormality diagnosis unit 7 that is connected to the vibration waveform collection unit 3 by a cable 5 and analyzes the vibration waveform data collected by the vibration waveform collection unit 3 to diagnose the presence or absence of an abnormality.
Each component will be described in detail.

<Vibration waveform collection section>
1, the vibration waveform collecting unit 3 includes a vibration sensor 13 consisting of a piezoelectric element 9 and a magnet 11 for placing the piezoelectric element 9 at a measurement location, an amplifier 15 for amplifying the signal output from the vibration sensor 13, and an A/D converter 17 for converting the analog data of the vibration amplified by the amplifier 15 into a digital signal. The vibration sensor 13 is attached by the magnet 11 near the output shaft of the drive unit 43 (position indicated by P in the figure).
The vibration waveform data acquired by the vibration waveform acquisition unit 3 is transmitted to the abnormality presence/absence diagnosis unit 7 via a cable 5 .

In this embodiment, an example is shown in which one vibration waveform acquisition unit 3 is used, but the number of vibration waveform acquisition units 3 is not limited to one in the present invention, and the use of multiple units is not excluded.

<Abnormality diagnosis section>
The abnormality diagnosis unit 7 is configured by devices and programs installed in a PC (personal computer).
As shown in FIG. 1, the abnormality diagnosis unit 7 includes a vibration waveform data storage means 19, a wavelet transform means 21, a wavelet coefficient time waveform acquisition means 23, an autocorrelation function calculation means 25, a correlation peak extraction means 27, a determination means 29, an input means 31 such as a keyboard, and a display means 33 such as a monitor.
The vibration waveform data storage means 19, wavelet transformation means 21, wavelet coefficient time waveform acquisition means 23, autocorrelation function calculation means 25, correlation peak extraction means 27, and judgment means 29 are realized by the CPU installed in the PC executing a program.
Each means will be described in detail below.

<<Vibration waveform data storage means>>
The vibration waveform data storage means 19 receives the signal transmitted from the vibration sensor unit, performs appropriate filtering on the vibration waveform data, and stores the data.

Wavelet transformation method
The wavelet transform means 21 performs wavelet transform on the collected vibration waveform to obtain the time variation of the wavelet coefficient for all frequencies.
The wavelet transform is a frequency analysis method used to see what frequencies occur with what amplitudes over time.
Here, the reason why wavelet transform is used in the present invention will be explained by taking as an example a sludge scraper that is the subject of diagnosis in this embodiment.
Sludge collectors are designed based on the amount of sludge to be treated, so there are different specifications for each facility, such as the size and structure of the facility, the types and dimensions of the mechanical parts used, etc. Therefore, if the specifications differ, the vibration frequency that occurs when an abnormality occurs will also differ.

The vibration frequency that occurs during this abnormality can be identified by conducting abnormality simulation tests or vibration excitation tests on individual sludge collectors, but conducting simulation tests on sludge collectors that are in operation is not realistic because it takes a considerable amount of time and effort.
However, even if the specifications of the sludge collector are different, it is true that if an abnormality such as wear occurs in the sprockets inside the collector tank, periodic vibrations in a certain frequency band will appear; it is just that the frequency band that occurs when an abnormality occurs differs depending on the specifications of the collector.
Therefore, by using wavelet transform, it is possible to calculate whether or not there is periodic abnormal vibration for each frequency for all frequencies in the frequency range to be analyzed, and it is possible to obtain information on whether or not there is periodic vibration. Therefore, even if the frequency band where the abnormality occurs is unknown, it is possible to determine whether or not there is an abnormality without specifying it.

<<Means for obtaining wavelet coefficient time waveform>>
The wavelet coefficient time waveform acquiring means 23 acquires the time waveform of the wavelet coefficient at each frequency at a predetermined interval, which is the change over time of the wavelet coefficient obtained by the wavelet transform.

<<Autocorrelation Function Calculation Means>>
The autocorrelation function calculation means 25 calculates the autocorrelation function for the time waveform of the wavelet coefficients of each frequency.

<<Means for extracting correlation peaks>>
The correlation peak extracting means 27 extracts an arbitrary number of peak values of the correlation in the autocorrelation function for each frequency.

《Judgment means》
The determining means 29 determines whether or not abnormal periodic vibration has occurred based on the relationship between the peak value extracted for each frequency and time.
A specific example of the determination method by the determination means 29 will be shown later in the description of the abnormality diagnosis method.

[Method for diagnosing abnormalities in mechanical equipment]
Next, a method for diagnosing the presence or absence of abnormality in mechanical equipment using the above-described abnormality diagnosis device 1 (hereinafter simply referred to as "abnormality diagnosis method") will be described.
As shown in FIG. 2, the abnormality diagnosis method according to this embodiment includes a vibration waveform acquisition step, a wavelet transform step, a wavelet coefficient time waveform acquisition step, an autocorrelation function calculation step, a correlation peak extraction step, and a judgment step.
Each step will be explained below.

<Vibration waveform collection process>
The vibration waveform acquisition step is a step of acquiring a vibration acceleration waveform for a predetermined period of time using a vibration sensor 13 installed in the target facility.
Specifically, the piezoelectric element 9 of the vibration waveform collecting unit 3 shown in FIG. 1 is attached to the driving device 43 of the sludge collector 41 by a magnet 11 to collect the vibration waveform.
The time for acquiring the vibration waveform data is a predetermined time that has been set in advance.
FIG. 3 shows an acceleration waveform obtained when acceleration is acquired as vibration data over a 300-second acquisition time.

<Wavelet transformation process>
The wavelet transform step is a step in which the wavelet transform means 21 performs wavelet transform on the collected vibration waveform to obtain the time change of the wavelet coefficient for all frequencies.

FIG. 4 shows an example of the change over time of the wavelet coefficient, where the vertical axis indicates frequency (Hz), the horizontal axis indicates time (sec), and the shade of gray indicates the strength of the wavelet coefficient.
FIG. 4(a) shows a case where the sprocket 47 is worn, and FIG. 4(b) shows a case where the sprocket 47 is in a normal state.

<Wavelet coefficient time waveform acquisition process>
The wavelet coefficient time waveform acquisition process is a process performed by the wavelet coefficient time waveform acquisition means 23, which acquires the time waveform of the wavelet coefficient at each frequency for each frequency at a predetermined interval from the time change of the wavelet coefficient obtained by the wavelet transform.
For example, the entire frequency band is divided into a predetermined number of parts (for example, 128 parts), and the time waveform of the wavelet coefficient is obtained for each divided frequency band (for example, every 14 Hz).

Figure 5 shows a time waveform graph of the wavelet coefficients of each frequency (a, b, c, ... z frequency) acquired by the wavelet coefficient time waveform acquisition means 23, with the vertical axis representing amplitude (m/ ^s2 ) and the horizontal axis representing time (sec).

<Autocorrelation function calculation step>
The autocorrelation function calculation step is a step in which the autocorrelation function calculation means 25 calculates the autocorrelation function for the time waveform of the wavelet coefficient of each frequency.
FIG. 6 shows an autocorrelation function, with the vertical axis representing the degree of autocorrelation and the horizontal axis representing time (sec).
By calculating the autocorrelation function, the periodicity of the intensity of the wavelet coefficients can be extracted.

<Correlation Peak Extraction Process>
The correlation peak extraction step is a step in which the correlation peak extraction means 27 extracts an arbitrary number of peak values of the correlation in the autocorrelation function for each frequency.
For example, in the example shown in FIG. 6, four points (indicated by dotted circles in the figure) with the largest peak values are extracted in the case of z Hz.

<Judgment process>
The determination step is a step in which the determination means 29 determines whether or not abnormal periodic vibration has occurred based on the relationship between the peak value extracted for each frequency and time.
As an example of a specific judgment method, a distribution diagram showing the relationship between the extracted peak value (correlation degree) and time is created, the center of gravity of the distribution diagram is calculated, and compared with the center of gravity of a normal state obtained in advance, and the presence or absence of an abnormality is judged based on the degree of deviation between the two.
Here, the reason why the presence or absence of an abnormality is determined based on the center of gravity of the distribution map will be explained.
The reason for calculating the center of gravity from the distribution map is that when the sludge collector's sprockets wear, the number of worn teeth, the shape of the wear, and the degree of wear can cause subtle differences in the meshing cycle between the sprocket and chain, resulting in a certain degree of variation in the distribution of peak values. However, even if there is variation in the meshing cycle, if there is an abnormality, peaks that do not appear under normal conditions will appear periodically in the autocorrelation function graph, causing the center of gravity of the distribution map to deviate from normal conditions. Utilizing this fact can simplify the determination of whether or not there is an abnormality.

FIG. 7 shows a distribution diagram in a state where an abnormality occurs in the sprocket 47 shown in FIG. 4(a), and the black circle in the diagram indicates the center of gravity of the distribution diagram.
FIG. 8 is a diagram in which the center of gravity of the distribution diagram obtained in the normal state shown in FIG. 4(b) is plotted on the distribution diagram shown in FIG.

As shown in Figure 8, the center of gravity during an abnormality is significantly different from the center of gravity during normal operation. A threshold value is set in advance to determine the degree of this deviation that is required to determine an abnormality, and an abnormality is determined if this threshold value is exceeded.

Here, a specific concept for setting the threshold value will be explained.
When a sprocket wears, periodic vibrations occur, and periodic, continuous peak components appear on the autocorrelation function graph. The period during which these peak components occur is based on the mathematically determined meshing period (measured in seconds) between the sprocket and chain, and peak components appear on the autocorrelation function graph as repetitive integer multiples (roughly five times) of that basic period. Therefore, since it was confirmed that it was possible to determine the presence or absence of an abnormality by limiting the analysis range to components up to five times the basic period, it was decided to take the center of gravity of multiple peak components up to five times the basic period.
As a result, it can be determined that a center of gravity that deviates approximately three times the meshing cycle of the sprocket and chain is in an abnormal state, so the threshold value can be set to a value approximately three times the meshing cycle of the sprocket and chain.

In the example shown in FIG. 8, the center of gravity in normal conditions is 32 [sec] on the time axis, and the threshold value is set to 80 [sec], which is about three times the normal value.
On the other hand, the center of gravity during an abnormality is 110 [sec], which is significantly greater than the threshold value of 80 [sec], so it is determined that an abnormality has occurred.

As described above, in this invention, the time waveforms of wavelet coefficients are acquired for all frequencies, and the presence or absence of abnormal periodic vibration is determined based on the time waveforms of the wavelet coefficients. This makes it possible to diagnose the presence or absence of abnormalities in mechanical equipment where the frequency band of the vibration waveform, etc., is unknown.

In the above description, the method for determining whether or not abnormal periodic vibration has occurred based on the relationship between the extracted peak value and time for each frequency involves calculating the center of gravity of a distribution diagram showing the relationship between the extracted peak value and time, comparing it with a previously determined center of gravity for normal conditions, and determining whether or not an abnormality has occurred based on the degree of deviation between the two.
However, the determination of the occurrence of abnormal periodic vibration in the present invention is not limited to this, and may be made by using the most frequent value or median value of the distribution showing the relationship between the extracted peak value and time.

REFERENCE SIGNS LIST 1 Abnormality diagnosis device 3 Vibration waveform sampling section 5 Cable 7 Abnormality diagnosis section 9 Piezoelectric element 11 Magnet 13 Vibration sensor 15 Amplifier 17 A/D converter 19 Vibration waveform data storage means 21 Wavelet transformation means 23 Wavelet coefficient time waveform acquisition means 25 Autocorrelation function calculation means 27 Correlation peak extraction means 29 Determination means 31 Input means 33 Display means 41 Sludge collector 43 Drive device 45 Endless chain 47 Sprocket (53, 54, 55, 57, 59, 61)
49 Flight 51 Settling tank 53 Drive device sprocket 54 Drive shaft drive sprocket 55, 57, 59, 61 Drive shaft driven sprocket

Claims (4)

A mechanical equipment abnormality diagnosis device that diagnoses the presence or absence of an abnormality in mechanical equipment,
The apparatus includes a vibration waveform acquisition unit that acquires vibration waveforms for a predetermined period of time using a vibration sensor installed in the target equipment, and an abnormality presence/absence diagnosis unit that analyzes the vibration waveform data acquired by the vibration waveform acquisition unit and diagnoses the presence or absence of an abnormality,
The abnormality diagnosis unit
wavelet transform means for performing wavelet transform on the collected vibration waveform to obtain time variations of wavelet coefficients for all frequencies;
a wavelet coefficient time waveform acquisition means for acquiring a time waveform of a wavelet coefficient at each frequency at a predetermined interval from the time change of the wavelet coefficient obtained by the wavelet transform;
an autocorrelation function calculation means for calculating an autocorrelation function for the time waveform of the wavelet coefficients of each frequency;
a correlation peak extraction means for extracting an arbitrary number of peak values of the correlation degree in the autocorrelation function for each frequency;
a determination means for determining whether or not abnormal periodic vibration has occurred based on the relationship between the peak value extracted for each frequency and time;
A diagnostic device for detecting abnormalities in mechanical equipment, comprising: The mechanical equipment abnormality diagnosis device described in claim 1, characterized in that the judgment means creates a distribution diagram showing the relationship between the extracted peak values and time, calculates the center of gravity of the distribution diagram, compares it with a predetermined normal center of gravity, and judges whether or not an abnormality exists based on the degree of deviation between the two. A method for diagnosing whether or not there is an abnormality in mechanical equipment, comprising:
a vibration waveform acquisition step of acquiring a vibration waveform for a predetermined period of time using a vibration sensor installed in the target facility;
a wavelet transform process for performing wavelet transform on the collected vibration waveform to obtain time variations of wavelet coefficients for all frequencies;
a wavelet coefficient time waveform acquisition step of acquiring a time waveform of the wavelet coefficient at each frequency at a predetermined interval from the time change of the wavelet coefficient obtained by the wavelet transform;
an autocorrelation function calculation step of calculating an autocorrelation function for the time waveform of the wavelet coefficients of each frequency;
a correlation peak extraction step of extracting an arbitrary number of peak values of the correlation degree in the autocorrelation function for each of the frequencies;
a determination step of determining whether or not abnormal periodic vibration has occurred based on the relationship between the peak value extracted for each frequency and time;
A method for diagnosing whether or not there is an abnormality in mechanical equipment, comprising: The method for diagnosing abnormalities in mechanical equipment described in claim 3, characterized in that the judgment step involves creating a distribution diagram showing the relationship between the extracted peak values and time, calculating the center of gravity of the distribution diagram, comparing it with a predetermined normal center of gravity, and judging the presence or absence of an abnormality based on the degree of deviation between the two.