Motor current analyzer, motor current analysis method, and program

Abstract

To provide a motor current analyzer capable of properly detecting abnormalities without increasing the time and effort, or the like, a motor current analysis method, and a program.SOLUTION: The motor current analyzer includes: a frequency analysis unit that performs frequency analysis for each period of time series of motor current values for multiple periods with a predetermined period as one cycle; a probability density distribution calculation unit that calculates the probability density distribution for each of multiple intervals set on a frequency axis based on frequency analysis results for multiple periods; and an anomaly calculation unit that calculates the degree of anomaly based on the obtained probability density distribution for each of the multiple intervals and reference probability density distribution.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present disclosure relates to an apparatus for analyzing motor current, a method for analyzing motor current, and a program.

Background Art

[0002] Patent Document 1 describes a diagnostic apparatus that determines an abnormality of a rotating machine based on the deviation between the distribution of each of a plurality of feature amounts respectively indicating the features of a current from the current waveform of the current measured during rotation of the rotating machine and the distribution of each of the plurality of feature amounts during normal operation of the rotating machine or the deviation of the multidimensional distribution of the plurality of feature amounts. According to the diagnostic apparatus described in Patent Document 1, for example, compared with the abnormality determination based on the deviation between the distribution of a single feature amount and the reference distribution, more comprehensive abnormality detection regarding the characteristics and types of abnormalities of the rotating machine becomes possible.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] As described above, according to the diagnostic apparatus described in Patent Document 1, compared with the abnormality determination based on the deviation between the distribution of a single feature amount and the reference distribution, the abnormality of the rotating machine can be detected more appropriately. However, in the diagnostic apparatus described in Patent Document 1, for example, verification of a plurality of feature amounts is required, so there is a problem that, depending on the case, it may take more labor and time, for example, during development, design, etc., than abnormality detection using a single feature amount.

[0005] The present disclosure has been made to solve the above problems, and an object thereof is to provide an apparatus for analyzing motor current, a method for analyzing motor current, and a program that can appropriately detect an abnormality without increasing labor or the like. [Means for solving the problem]

[0006] To solve the above problems, the motor current analyzer according to this disclosure comprises: a frequency analysis unit that performs frequency analysis on a time series of motor current values, with each predetermined period being one cycle, for multiple cycles; a probability density distribution calculation unit that calculates a probability density distribution for multiple intervals set on the frequency axis based on the frequency analysis results for the multiple cycles; and an abnormality calculation unit that calculates the degree of abnormality based on the calculated probability density distribution for each of the multiple intervals and a reference probability density distribution.

[0007] To solve the above problems, the motor current analysis device according to this disclosure comprises a probability density distribution calculation unit that calculates a probability density distribution for each of several intervals set on the time axis of a predetermined period based on multiple time series of motor current values, each of which has a predetermined period as one cycle, and an abnormality calculation unit that calculates the degree of abnormality based on the calculated probability density distribution for each of the multiple intervals and a reference probability density distribution.

[0008] The motor current analysis method according to this disclosure includes the steps of: performing frequency analysis on a time series of motor current values, with each predetermined period being one cycle, for multiple cycles; determining a probability density distribution for multiple intervals set on the frequency axis based on the frequency analysis results for the multiple cycles; and calculating the degree of abnormality based on the determined probability density distribution for each of the multiple intervals and a reference probability density distribution.

[0009] The motor current analysis method according to this disclosure includes the steps of: determining a probability density distribution for each of several intervals set on the time axis of a predetermined period based on multiple time series of motor current values, each of which has a predetermined period as one cycle; and calculating the degree of abnormality based on the probability density distribution for each of the multiple intervals obtained and a reference probability density distribution.

[0010] The program relating to this disclosure includes the steps of: performing frequency analysis on a time series of motor current values, with each period defined as a predetermined period, for multiple periods; and determining a probability density distribution for each of multiple intervals set on the frequency axis based on the frequency analysis results for the multiple periods. The computer is instructed to perform the step of calculating the degree of anomaly based on the probability density distributions for each of the multiple intervals obtained and a reference probability density distribution.

[0011] The program relating to this disclosure causes a computer to perform the following steps: determine a probability density distribution for each of several intervals set on the time axis of a predetermined period based on multiple time series of motor current values, each of which has a predetermined period as one cycle; and calculate the degree of abnormality based on the determined probability density distribution for each of the multiple intervals and a reference probability density distribution. [Effects of the Invention]

[0012] According to the motor current analyzer, motor current analyzer, and program of this disclosure, abnormalities can be appropriately detected without increasing the amount of work involved. [Brief explanation of the drawing]

[0013] [Figure 1] This is a schematic plan view showing the general configuration of a filling apparatus according to an embodiment of the present disclosure. [Figure 2] This figure shows a schematic plan view and a side view illustrating the general configuration of the drive system of a filling apparatus according to an embodiment of the present disclosure. [Figure 3] This is a block diagram showing a schematic configuration of the control system of a filling apparatus according to an embodiment of the present disclosure. [Figure 4] This block diagram shows the schematic configuration of a data storage and analysis PC according to an embodiment of this disclosure. [Figure 5] This is a block diagram showing the functional configuration of the abnormality determination processing unit according to the first embodiment of this disclosure. [Figure 6] This flowchart shows an example of the operation of a data storage and analysis PC according to the first embodiment of this disclosure. [Figure 7]This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the first embodiment of the present disclosure. [Figure 8] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the first embodiment of the present disclosure. [Figure 9] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the first embodiment of the present disclosure. [Figure 10] This is a block diagram showing a functional configuration of an abnormality determination processing unit according to the second embodiment of the present disclosure. [Figure 11] This is a flowchart showing an operation example of a PC for data storage and analysis according to the second embodiment of the present disclosure. [Figure 12] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the second embodiment of the present disclosure. [Figure 13] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the second embodiment of the present disclosure. [Figure 14] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the second embodiment of the present disclosure. [Figure 15] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the third embodiment of the present disclosure. [Figure 16] This is a schematic diagram for explaining an operation example of a PC for data storage and analysis according to the third embodiment of the present disclosure. [Figure 17] This is a schematic block diagram showing a configuration of a computer according to at least one embodiment.

Mode for Carrying Out the Invention

[0014] <First Embodiment> (Configuration of the First Embodiment) Hereinafter, a motor current analysis apparatus, a motor current analysis method, and a program according to the first embodiment of this disclosure will be described with reference to Figures 1 to 9. Figure 1 is a schematic plan view showing the general configuration of a filling apparatus according to an embodiment of this disclosure (common to the first to third embodiments). Figure 2 is a diagram showing a schematic plan view A and a side view B of the drive system of the filling apparatus according to an embodiment of this disclosure. Figure 3 is a block diagram showing the schematic configuration of the drive system of the filling apparatus according to an embodiment of this disclosure. Figure 4 is a block diagram showing the schematic configuration of a data storage and analysis PC according to an embodiment of this disclosure. Figure 5 is a block diagram showing the functional configuration of an abnormality determination processing unit according to the first embodiment of this disclosure. Figure 6 is a flowchart showing an example of operation of the data storage and analysis PC according to the first embodiment of this disclosure. Figures 7 to 9 are schematic diagrams for explaining an example of operation of the data storage and analysis PC according to the first embodiment of this disclosure. In each figure, the same or corresponding components are used with the same reference numerals and explanations are omitted as appropriate.

[0015] As shown in Figure 1, the filling system 1 according to the embodiment of this disclosure includes a filling device 100 for performing processes such as filling product liquid into containers such as bottles, cans, and plastic bottles. The filling device 100 includes rotating bodies 111 to 119 called star wheels that are rotated by a drive source such as an AC motor, a filling machine 121 for filling product liquid into containers, and a capping device 131 for sealing containers. Hereinafter, the rotating bodies 111 to 119 will be collectively referred to as rotating body 110. Also, the rotating body 110, the filling machine 121, and the capping device 131 will be collectively referred to as rotating body 110, etc.

[0016] The rotating bodies 110, etc., are equipped with multiple gripping devices called grippers (not shown) on their outer circumference, etc. The rotating bodies 110, etc., grip a part of the container, such as the body or neck, with the gripping devices and transport the container while rotating in the direction indicated by the arrow. The rotating bodies 110, etc., are installed adjacent to each other, and containers are sequentially passed from the upstream rotating body 110, etc. to the downstream rotating body 110, while processes such as beverage filling, sealing, and labeling of the containers are carried out. In the example shown in Figure 1, containers (not shown) that have been molded by the container molding device 200 are supplied to the filling device 100, and the containers are transported in the order of rotating bodies 111-116, filling machine 121, rotating body 117, capping device 131, and rotating bodies 118-119, and the containers are discharged from rotating body 119.

[0017] As described above, each rotating body 110 is equipped with multiple gripping devices around it. For example, when a container is being transported and processed on the rotating body 111, the gripping devices of the rotating body 111 hold the container. However, when the processing is finished and the container is handed over to the downstream rotating body 112, the gripping devices of the rotating body 111 release the container. The gripping devices of the receiving rotating body 112 receive the container with their open state, and once received, close the gripping devices to grasp the container. The transfer and receipt of containers are performed synchronously on each rotating body 110. However, the radial dimensions, i.e., diameters, of the rotating bodies 110 may not be the same. Therefore, the number of gripping devices on each rotating body 110 may be the same or different. For example, the rotating bodies 110 are driven in a state where no container is gripped by the gripping devices at the start of processing (referred to as an unloaded state), and after a predetermined time has elapsed, they are driven in a state where, for example, all gripping devices are gripping a container (referred to as an actual load state).

[0018] As shown in Figure 2, each rotating body 110, etc., is connected to a drive motor 150 via a drive shaft 140. Furthermore, as shown in Figure 3, each drive motor 150 is driven by a plurality of motor control devices 12 having inverters, etc. Each motor control device 12 is controlled by a filling machine control device 11. The filling machine control device 11 is configured using, for example, a PLC (Programmable Logic Controller), and receives sensor signals, etc., representing the detection results of multiple sensors equipped on the motor control devices 12, drive motors 150, etc., and controls the motor control devices 12, etc. The data acquisition unit 13 is configured using a PLC, etc., and collects and records sensor signals, control signals, etc., generated by the filling machine control device 11, etc. from the filling machine control device 11. In this embodiment, the sensor signals, control signals, etc., are collectively referred to as process data. Furthermore, in this embodiment, the process data includes a time series of the motor current (drive current) of each drive motor 150 (for example, time series data representing the instantaneous value of the current measured at a predetermined sampling period; also called a current-time waveform), a time series of the rotation angle of each drive motor 150, and the state of the load on the rotating body 110, etc. (for example, a signal indicating whether each gripping device is gripping a container or not).

[0019] The data storage and analysis PC (personal computer) 14 receives predetermined process data from the data acquisition unit 13, stores it in an internal memory device, and analyzes, for example, the degree of abnormality (degree of abnormality) of motor current based on the process data according to the operator's instructions. The data storage and analysis PC 14 is equipped with a processor, input devices such as a keyboard and pointing device, output devices such as a display device, and memory devices as internal or peripheral devices, and its functional configuration, consisting of a combination of these hardware components and software such as programs executed by the processor, includes the following parts as shown in Figure 4. Specifically, the data storage and analysis PC 14 includes a data preprocessing unit 141, an abnormality determination processing unit 142, an abnormality detection result output unit 143, a data storage unit 144, and a model creation processing unit 145. In this embodiment, the data storage and analysis PC 14 is one example of the configuration of the motor current analysis device according to this disclosure.

[0020] The data preprocessing unit 141 performs preprocessing on the process data D1 received from the data acquisition unit 13, such as noise reduction and interpolation, and stores it in the data storage unit 144. The data preprocessing unit 141 also outputs the preprocessed process data to the anomaly detection unit 142.

[0021] The model creation processing unit 145 creates a model that simulates the motor current of each drive motor 150, for example, based on process data stored in the data storage unit 144. Using the model, it calculates the time series of motor currents under normal load and no-load conditions, and further calculates the probability density distribution of predetermined features based on the motor current time series as reference data D2. Note that the method of calculating (creating) reference data D2 is not limited to this example; for example, it may be calculated from the actual process data D1 itself under normal conditions. In this embodiment, the probability density distribution is the distribution of a random variable defined by a probability density function and a probability mass function, and is represented by a normal distribution, etc. Reference data D2 can be defined, for example, by the type of distribution such as a normal distribution, and the mean and standard deviation (or variance). Reference data D2 can also be calculated for each interval (dimension) described later. The features in this embodiment will be described later.

[0022] The anomaly detection processing unit 142 calculates the degree of anomaly based on the probability density distribution of a predetermined feature quantity based on the time series of motor current values ​​included in process data D1 and reference data D2 (reference probability density distribution). It then compares the calculated degree of anomaly with a predetermined degree of anomaly judgment value to determine whether the degree of anomaly is greater than or equal to the degree of anomaly judgment value. Here, the degree of anomaly is an index that represents the magnitude of the statistical difference (distance) from the reference data D2.

[0023] The anomaly detection result output unit 143 outputs (for example, displays) a statement to the effect of the anomaly determination processing unit 142, or the value of the anomaly, if the anomaly determination processing unit 142 determines that the degree of anomaly is equal to or greater than the degree of anomaly determination value.

[0024] Figure 5 shows an example of the configuration of the anomaly detection processing unit 142 shown in Figure 4. The anomaly detection processing unit 142 shown in Figure 4 has the following functional configuration: a data acquisition unit 1421, a frequency analysis unit 1422, an interval extraction unit 1423, a feature calculation unit 1424, a probability density distribution calculation unit 1425, an anomaly degree calculation unit 1426, and an anomaly degree determination unit 1427.

[0025] The data acquisition unit 1421 acquires multiple time series of motor current values ​​from the data preprocessing unit 141 (or data storage unit 144), with each period being a predetermined duration. The frequency analysis unit 1422 performs frequency analysis on the time series of motor current values ​​for each of the multiple periods and calculates a frequency spectrum as a result of the frequency analysis. Figure 7 shows an example of the time series of motor current values ​​(current-time waveform) for one period acquired by the data acquisition unit 1421 and the frequency analysis result for one period calculated by the frequency analysis unit 1422. Figure 7 shows the current-time waveform with time on the horizontal axis and the instantaneous value of current on the vertical axis, and the frequency analysis result with frequency on the horizontal axis and gain (spectral intensity or component level) on the vertical axis. In the example shown in Figure 7, the predetermined period (1 period) T1 is the time from time t0 to time t1. The frequency analysis result shown in Figure 7 is the power spectrum obtained as a result of FFT (Fast Fourier Transform) analysis with this 1 period T1 as the time window. The start time t0 of the predetermined period T1 can be set to, for example, the position of the drive motor 150 (0 degrees), and the end time t1 can correspond to the position after one rotation (360 degrees) (or a position after multiple rotations or a predetermined rotation less than one rotation). Also, one period T1 may or may not overlap with the next period T1. The frequency analysis unit 1422 calculates frequency analysis results for multiple periods as shown in Figure 7.

[0026] The interval extraction unit 1423 extracts data sets of gain values ​​for each of the multiple intervals set on the frequency axis in the frequency analysis results. In the example shown in Figure 7, two intervals are set: the first interval is the frequency band FB1 from frequency f1 to frequency f2, and the second interval is the frequency band FB2 from frequency f2 to frequency f3. In this case, the interval extraction unit 1423 extracts data sets of gain values ​​for frequency band FB1 and data sets of gain values ​​for frequency band FB2. Note that the number of intervals is not limited to two, but may be three or more. Also, each interval may be continuous, discontinuous, or contain overlapping frequency bands.

[0027] The feature calculation unit 1424 calculates features for each interval from which the interval extraction unit 1423 has extracted the data set. In this embodiment, the features are assumed to be the average value of the gain or the peak value multiplier for each interval. However, other features may be used in other embodiments. The average value of the gain for each interval is, for example, the average value of each gain value belonging to frequency band FB1 (this is called average gain 1), and the average value of each gain value belonging to frequency band FB2 (this is called average gain 2). The peak value multiplier is the multiplier of the peak value (maximum gain) to the average value of the gain. The peak value multiplier is, for example, the ratio of the maximum gain belonging to frequency band FB1 to the average value of each gain belonging to frequency band FB1, and the ratio of the maximum gain belonging to frequency band FB2 to the average value of each gain belonging to frequency band FB2. The feature calculation unit 1424 calculates features for each period for the number of intervals set for the frequency analysis results for multiple periods calculated by the frequency analysis unit 1422.

[0028] The probability density distribution calculation unit 1425 calculates the probability density distribution for each of the multiple intervals set on the frequency axis based on the frequency analysis results for multiple periods. Figure 8 shows an example of the probability density distribution calculation by the probability density distribution calculation unit 1425. The example shown in Figure 8 shows an example of the calculation results for the probability density distribution of the average gain 1 for multiple periods (a distribution with average gain 1 as the random variable) and the probability density distribution of the average gain 2 for multiple periods (a distribution with average gain 2 as the random variable), with two intervals set as shown in Figure 7. In this case, the calculation example shown in Figure 8 is a two-dimensional probability density distribution corresponding to 2 intervals. Note that if the feature quantity is the peak value multiplier, in Figure 8, average gain 1 will be read as peak value multiplier 1 (peak value multiplier based on each gain belonging to the frequency band FB1), and average gain 2 will be read as peak value multiplier 2 (peak value multiplier based on each gain belonging to the frequency band FB2). In this embodiment, the probability density distribution calculation unit 1425 calculates the probability density distribution for each of the multiple intervals using the average gain or peak value multiplier for each interval obtained as a result of the frequency analysis as the random variable.

[0029] The anomaly calculation unit 1426 calculates the anomaly score using, for example, the following formula (1), based on the probability density distributions for each of the multiple intervals obtained by the probability density distribution calculation unit 1425 and a reference probability density distribution (reference data D2).

[0030]

number

[0031] In equation (1), the "normal distribution" is the probability density distribution of the reference data D2, and the "detection target distribution" is the probability density distribution of the detection target calculated by the probability density distribution calculation unit 1425 based on the process data D1.

[0032] When calculating the degree of anomaly, the degree of anomaly calculation unit 1426 calculates the degree of anomaly based on, for example, the mean and variance or standard deviation of a one-dimensional probability density distribution obtained by dimensionality reduction of the probability density distributions for multiple intervals as a multidimensional probability density distribution, and the mean and variance or standard deviation of a reference probability density distribution. Alternatively, the degree of anomaly calculation unit 14265 calculates multiple degrees of anomaly for each dimension based on the mean and variance or standard deviation of the probability density distribution for each dimension, and the mean and variance or standard deviation of a reference probability density distribution, using the probability density distributions for multiple intervals as a multidimensional probability density distribution.

[0033] Figure 9 shows an example of converting the two-dimensional probability density distribution shown in Figure 8 into a one-dimensional probability density distribution through dimensionality reduction. Dimensionality reduction is the process of finding common features in multidimensional data and ignoring them to reduce the data's dimension, for example, from two dimensions to one. While dimensionality reduction is another method for reducing dimensionality (a method for reducing the data's dimension from two to one while minimizing information loss), dimensionality reduction can also be used when converting a multidimensional probability density distribution to a one-dimensional one.

[0034] The abnormality determination unit 1427 compares the abnormality level calculated by the abnormality calculation unit 1426 with a predetermined abnormality determination value, and determines that there is an abnormality if the abnormality level is equal to or greater than the abnormality determination value. The abnormality determination value can be calculated, for example, by using a model created by the model creation processing unit 145 to determine a value within an acceptable range for variations in the operation of motors, etc., or by experimentally creating a state close to an abnormal state and setting it based on measured values, etc.

[0035] (Example of operation of the first embodiment) Next, an example of operation of this embodiment will be described with reference to Figure 6. The process shown in Figure 6 is an example of abnormality determination processing for one drive motor 150 performed by the abnormality determination processing unit 142. When the process shown in Figure 6 is started, first the data acquisition unit 13 performs current measurement for multiple cycles at regular intervals (for example, every cycle T1 shown in Figure 7) (step S11). Next, the data acquisition unit 1421 acquires pre-processed process data D1, and the frequency analysis unit 1422 performs FFT analysis to calculate the frequency spectrum for multiple cycles (step S12). Next, for each interval extracted by the interval extraction unit 1423 (by specific frequency band), the feature calculation unit 1424 calculates the average gain or peak value multiplier for multiple cycles (step S13). Next, the probability density distribution calculation unit 1425 calculates the average and standard deviation of the average gain or peak value multiplier for multiple cycles for each dimension (step S15).

[0036] Next, the anomaly calculation unit 1426 calculates the difference between the mean and standard deviation of the reference data (anomaly score by frequency analysis) for each dimension based on equation (1) (step S15). Next, the anomaly determination unit 1427 determines for each dimension whether the anomaly score is greater than or equal to the anomaly determination value (step S16). If the anomaly score is greater than or equal to the anomaly determination value (step S16:Y), it determines that there is an anomaly (step S17), sets an anomaly flag (based on this flag, the anomaly detection result output unit 143 outputs that an anomaly has been detected), and terminates the process shown in Figure 6. If the anomaly score is not greater than or equal to the anomaly determination value (step S16:N), the anomaly determination unit 1427 determines that there is no anomaly (step S19), and terminates the process shown in Figure 6.

[0037] (Effects and workings of the first embodiment) According to the first embodiment, the data storage and analysis PC 14 (motor current analysis device) includes a frequency analysis unit 1422 that performs frequency analysis on multiple periods of a time series of motor current values ​​with a predetermined period (period T1) as one period, a probability density distribution calculation unit 1425 that calculates a probability density distribution for multiple intervals set on the frequency axis based on the frequency analysis results for multiple periods, and an anomaly degree calculation unit 1426 that calculates the anomaly degree based on the calculated probability density distribution for multiple intervals and a reference probability density distribution. With this configuration, for example, by calculating multiple probability density distributions for each interval for a single feature, the anomaly degree can be calculated based on a multidimensional probability density distribution. Therefore, by appropriately setting the intervals, anomalies can be detected more appropriately than anomaly detection based on a one-dimensional probability density distribution that does not divide the interval based on a single feature. In addition, since the feature can be a single feature, the effort and other factors can be easily reduced compared to the case where multiple types of feature quantities are used. Therefore, according to this embodiment, anomalies can be detected appropriately without increasing the effort and other factors.

[0038] Furthermore, as in this embodiment, by comparing probability density distributions, it becomes easy to detect anomalies such as changes over time (e.g., the mean gradually shifting, or the variability increasing).

[0039] <Second Embodiment> (Configuration of the second embodiment) Next, a motor current analysis apparatus, a motor current analysis method, and a program according to the second embodiment of this disclosure will be described with reference to Figures 10 to 14. Figure 10 is a block diagram showing the functional configuration of the abnormality determination processing unit according to the second embodiment of this disclosure. Figure 11 is a flowchart showing an example of operation of the data storage and analysis PC according to the second embodiment of this disclosure. Figures 12 to 14 are schematic diagrams illustrating an example of operation of the data storage and analysis PC according to the second embodiment of this disclosure.

[0040] The second embodiment is basically the same as the first embodiment in terms of the configuration shown in Figures 1 to 4. However, in the second embodiment, the functional configuration of the anomaly determination processing unit 142 shown in Figure 4 is partially different. As shown in Figure 10, the anomaly determination processing unit 142a of the second embodiment (the configuration corresponding to the anomaly determination processing unit 142 of the first embodiment shown in Figure 5) does not have the frequency analysis unit 1422 and feature calculation unit 1424 that are included in the anomaly determination processing unit 142 shown in Figure 5, and the configuration and operation of the interval extraction unit 1423a (the configuration corresponding to the interval extraction unit 1423) and the probability density distribution calculation unit 1425a (the configuration corresponding to the probability density distribution calculation unit 1425) are partially different. The data acquisition unit 1421, the anomaly degree calculation unit 1426, and the anomaly degree determination unit 1427 are basically the same in the first and second embodiments.

[0041] The data acquisition unit 1421 acquires multiple time series of motor current values ​​from the data preprocessing unit 141 (or data storage unit 144), with each period being a predetermined duration. The interval extraction unit 1423a extracts data sets of current values ​​for each of multiple intervals set on the time axis of one period of the motor current value time series, with each period being a predetermined duration. In the example shown in Figure 12, in the current time waveform with the horizontal axis representing time and the vertical axis representing the instantaneous value of current, two intervals are set: the period T10 from time t10 to time t13 is defined as one period, the period T11 from time t10 to time t11 is defined as the first interval, and the period T12 from time t12 to time t13 is defined as the second interval. In this case, the interval extraction unit 1423a extracts data sets of current values ​​within period T11 and data sets of current values ​​within period T12. Note that the number of intervals is not limited to two, but may be three or more. Furthermore, each section may be continuous, discontinuous, or overlapping. The start time t10 of the predetermined period T10 may be, for example, the position of the drive motor 150 (0 degrees), and the end time t13 may correspond to the position after one rotation (360 degrees) (or a position after multiple rotations or a predetermined rotation of less than one rotation).

[0042] The probability density distribution calculation unit 1425a calculates the probability density distribution for multiple intervals set on the time axis of one period, based on multiple periods of the time series of motor current values, with a predetermined period T10 as one period. In this case, the probability density distribution calculation unit 1425a calculates the probability density distribution for multiple intervals, using the time series of motor current values ​​as a random variable. Figure 13 shows an example of the probability density distribution calculation by the probability density distribution calculation unit 1425a. The example shown in Figure 13 shows an example of the calculation results of the probability density distribution of current values ​​(each sample value) for multiple periods (distribution with current value 1 as the random variable) and the probability density distribution of current values ​​(each sample value) for multiple periods (distribution with current value 2 as the random variable), with two intervals set as shown in Figure 12. In this case, the calculation example shown in Figure 13 is a two-dimensional probability density distribution corresponding to 2 intervals. Figure 14 also shows an example of converting the two-dimensional probability density distribution shown in Figure 12 into a one-dimensional probability density distribution by dimensionality reduction. Furthermore, as in the first embodiment, the dimensionality may be reduced using a dimensionality reduction method.

[0043] (Example of operation of the second embodiment) Next, an example of operation of this embodiment will be described with reference to Figure 11. The process shown in Figure 11 is an example of abnormality determination processing for one drive motor 150 performed by the abnormality determination processing unit 142a shown in Figure 10. When the process shown in Figure 11 is started, first the data acquisition unit 13 performs current measurement for multiple cycles at regular intervals (for example, every cycle T10 shown in Figure 12) (step S21). Next, the data acquisition unit 1421 acquires pre-processed process data D1, the interval extraction unit 1423a performs phase alignment of the current time waveforms for multiple cycles (step S22), divides the phase-aligned current waveforms into specific intervals, and the probability density distribution calculation unit 1425a performs dimensionality reduction (step S23). Next, the probability density distribution calculation unit 1425a calculates the average and standard deviation of the current values ​​for multiple cycles (step S24).

[0044] Next, the abnormality calculation unit 1426 calculates the difference between the mean and standard deviation of the reference data (abnormality based on the current waveform) based on equation (1) (step S25). Next, the abnormality determination unit 1427 determines whether the abnormality is greater than or equal to the abnormality determination value (step S26). If the abnormality is greater than or equal to the abnormality determination value (step S26:Y), it determines that there is an abnormality (step S27), sets an abnormality flag (based on this flag, the abnormality detection result output unit 143 outputs that an abnormality has been detected), and terminates the process shown in Figure 11. If the abnormality is not greater than or equal to the abnormality determination value (step S26:N), the abnormality determination unit 1427 determines that there is no abnormality (step S29), and terminates the process shown in Figure 11.

[0045] (Effects and workings of the second embodiment) According to the second embodiment, the data storage and analysis PC 14 (motor current analysis device) includes a probability density distribution calculation unit 1425a that calculates a probability density distribution for each of several intervals set on the time axis of one period based on multiple periods of the time series of motor current values, with each period (period T10) being one period, and an anomaly degree calculation unit 1426 that calculates the degree of anomaly based on the calculated probability density distribution for each of the multiple intervals and a reference probability density distribution. With this configuration, for example, multiple probability density distributions can be calculated for each interval for a single feature quantity, and the degree of anomaly can be calculated based on a multidimensional probability density distribution. Therefore, by appropriately setting the intervals, anomalies can be detected more appropriately than, for example, anomaly detection based on a one-dimensional probability density distribution based on a single feature quantity. In addition, since the feature quantity can be a single feature quantity, the effort and other factors can be easily reduced compared to the case where multiple types of feature quantities are used. Therefore, according to this embodiment, anomalies can be detected appropriately without increasing the effort and other factors.

[0046] Furthermore, according to this embodiment, it is possible to capture changes in current that are difficult to capture using frequency alone.

[0047] <Third Embodiment> (Configuration of the third embodiment) Next, a motor current analysis apparatus, a motor current analysis method, and a program according to the third embodiment of this disclosure will be described with reference to Figures 15 and 16. Figures 15 and 16 are schematic diagrams illustrating an example of the operation of a data storage and analysis PC according to the third embodiment of this disclosure. The configuration of the third embodiment is basically the same as that of the first or second embodiment, and the third embodiment can be configured by changing a part of the configuration of the first or second embodiment. In the third embodiment, the time series of motor current values ​​is measured under multiple different load conditions (actual load condition, no-load condition, etc. (or intermediate load condition) as described above), the degree of abnormality is calculated for each load condition, and the degree of abnormality determination unit (configuration corresponding to the degree of abnormality determination unit 1427 of the first and second embodiments) estimates the nature of the abnormality by comparing multiple degrees of abnormality for each interval and for each of the multiple load conditions with a predetermined degree of abnormality determination value.

[0048] For example, when modifying a part of the configuration of the first embodiment, as shown in Figure 15, F1±ΔF1, F2±ΔF2, F3±ΔF3, and F4±ΔF4 are set as notable frequency bands corresponding to the intervals. For each of these frequency bands F1±ΔF1, F2±ΔF2, F3±ΔF3, and F4±ΔF4, the probability density distribution of the average value or peak value multiplier of the gain over multiple periods is determined, and the degree of abnormality is calculated based on the obtained probability density distribution and the probability density distribution of the reference data D2, and compared with a predetermined degree of abnormality judgment value. In the example shown in Figure 15, if an abnormality is found in frequency band F1±ΔF1 under actual load, abnormality (1) is estimated as the nature of the abnormality. Also, if an abnormality is found in frequency band F2±ΔF2 or frequency band F3±ΔF3 under no load, abnormality (2) or abnormality (3) is estimated as the nature of the abnormality. Furthermore, if there is an abnormality under actual load but no abnormality under no load in the frequency band F4±ΔF4, abnormality (4) is estimated, and if there is no abnormality under actual load but there is an abnormality under no load, abnormality (5) is estimated. Abnormalities (1) to (5) include, for example, bearing abnormalities, abnormalities in the transfer by the gripping device, etc. ΔF1 to ΔF4 are values ​​that represent the error range with respect to the center frequencies F1 to F4.

[0049] Furthermore, if, for example, part of the configuration of the second embodiment is changed, as shown in Figure 16, DEG1±ΔDEG1, DEG2±DEG2, and DEG3±DEG3 are set as notable phases corresponding to the intervals. For each of these phases DEG1±ΔDEG1, DEG2±DEG2, and DEG3±DEG3, the probability density distribution of current values ​​for multiple periods is determined, and the degree of abnormality is calculated based on the determined probability density distribution and the probability density distribution of the reference data D2, and compared with a predetermined degree of abnormality judgment value. In the example shown in Figure 16, if there is an abnormality in phase DEG1±ΔDEG1 under actual load, abnormality (6) is estimated as the nature of the abnormality. Also, if there is an abnormality in phase DEG2±ΔDEG2 under no load, abnormality (7) is estimated as the nature of the abnormality. Also, if there is an abnormality in phase DEG3±ΔDEG3 under actual load but no abnormality under no load, abnormality (8) is estimated, and if there is no abnormality under actual load but there is an abnormality under no load, abnormality (9) is estimated. Abnormalities (6) to (9) include, for example, bearing abnormalities, abnormalities in the transfer by the gripping device, etc. ΔDEG1 to ΔDEG3 are values ​​that represent the error range for the center angles DEG1 to DEG3.

[0050] (Effects and mechanisms of the third embodiment) According to this embodiment, it is possible to estimate what kind of abnormality is occurring based on the detected abnormality.

[0051] (Other embodiments) Although embodiments of this disclosure have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and may include design changes, etc., that do not depart from the gist of this disclosure. In each of the above embodiments, the calculation of abnormality based on motor current for the filling system 1 or filling device 100 was illustrated as an example, but the embodiments of this disclosure are not limited to this example. It can be applied to motors in general, such as AC motors and DC motors (electric motors or generators and rotating machines that become electric motors).

[0052] <Computer Configuration> Figure 17 is a schematic block diagram showing the configuration of a computer according to at least one embodiment. The computer 90 includes a processor 91, main memory 92, storage 93, and an interface 94. The aforementioned data storage and analysis PC 14 is implemented in the computer 90. The operation of each processing unit described above is stored in storage 93 in the form of a program. The processor 91 reads the program from storage 93, loads it into main memory 92, and executes the above processing according to the program. The processor 91 also allocates memory areas in main memory 92 corresponding to each of the storage units described above, according to the program.

[0053] The program may be for implementing some of the functions that the computer 90 is to perform. For example, the program may perform functions in combination with other programs already stored in storage, or in combination with other programs implemented in other devices. In other embodiments, the computer may include a custom LSI (Large Scale Integrated Circuit) such as a PLD (Programmable Logic Device) in addition to, or instead of, the above configuration. Examples of PLDs include PAL (Programmable Array Logic), GAL (Generic Array Logic), CPLD (Complex Programmable Logic Device), FPGA (Field Programmable Gate Array), etc. In this case, some or all of the functions implemented by the processor may be implemented by the integrated circuit.

[0054] Examples of storage 93 include HDDs (Hard Disk Drives), SSDs (Solid State Drives), magnetic disks, magneto-optical disks, CD-ROMs (Compact Disc Read Only Memory), DVD-ROMs (Digital Versatile Disc Read Only Memory), and semiconductor memory. Storage 93 may be an internal medium directly connected to the bus of the computer 90, or an external medium connected to the computer 90 via an interface 94 or a communication line. Furthermore, if this program is distributed to the computer 90 via a communication line, the computer 90 that receives the program may expand it into main memory 92 and execute the above processing. In at least one embodiment, storage 93 is a tangible storage medium that is not temporary.

[0055] <Note> The data storage and analysis PC14 (motor current analysis device) described in each embodiment can be understood, for example, as follows.

[0056] (1) The data storage and analysis PC 14 (motor current analysis device) according to the first embodiment includes a frequency analysis unit 1422 that performs frequency analysis on a time series of motor current values ​​with a predetermined period as one cycle for multiple cycles, for each of the said cycles, a probability density distribution calculation unit 1425 that calculates a probability density distribution for each of the multiple intervals set on the frequency axis based on the frequency analysis results for the said multiple cycles, and an abnormality calculation unit 1426 that calculates the degree of abnormality based on the calculated probability density distribution for each of the multiple intervals and a reference probability density distribution. According to this embodiment and each of the following embodiments, abnormalities can be appropriately detected without increasing the amount of work involved.

[0057] (2) The data storage and analysis PC 14 (motor current analysis device) according to the first embodiment is the data storage and analysis PC 14 of (1), wherein the probability density distribution calculation unit 1425 calculates the probability density distribution for each of the multiple intervals using the average value or peak value multiplier of the gain for each interval obtained as the frequency analysis result as a random variable.

[0058] (3) The data storage and analysis PC 14 (motor current analysis device) according to the third embodiment includes a probability density distribution calculation unit 1425a that calculates a probability density distribution for each of the multiple intervals set on the time axis of one cycle based on multiple cycles of the motor current value time series with a predetermined period as one cycle, and an abnormality calculation unit 1426 that calculates the degree of abnormality based on the calculated probability density distribution for each of the multiple intervals and a reference probability density distribution. According to this embodiment and each of the following embodiments, abnormalities can be appropriately detected without increasing the amount of work involved.

[0059] (4) The data storage and analysis PC 14 (motor current analysis device) according to the fourth embodiment is the data storage and analysis PC 14 of (3), wherein the probability density distribution calculation unit 1425a calculates the probability density distribution for each of the multiple intervals using the time series of the motor current value as a random variable.

[0060] (5) The data storage and analysis PC 14 (motor current analysis device) according to the fifth embodiment is the data storage and analysis PC 14 according to (1) to (4), wherein the time series of the motor current values ​​is measured under multiple different load conditions, the degree of abnormality is calculated for each load condition, and the abnormality determination unit (1427) further comprises an abnormality determination unit that estimates the nature of the abnormality by comparing multiple abnormality values ​​for each interval and for each of the multiple load conditions with a predetermined abnormality determination value.

[0061] (6) The data storage and analysis PC 14 (motor current analysis device) according to the sixth embodiment is the data storage and analysis PC 14 according to (1) to (5), wherein the abnormality calculation unit 1426 calculates the abnormality based on the mean and variance or standard deviation of a one-dimensional probability density distribution obtained by dimensionality reduction of the multidimensional probability density distribution obtained by treating the probability density distribution for each of the multiple intervals as a multidimensional probability density distribution, and the mean and variance or standard deviation of the reference probability density distribution.

[0062] (7) The data storage and analysis PC 14 (motor current analysis device) according to the seventh embodiment is the data storage and analysis PC 14 of (1) to (5), wherein the abnormality calculation unit 1426 calculates a plurality of abnormalities based on the mean and variance or standard deviation of the probability density distribution for each of the plurality of intervals as a multidimensional probability density distribution and the mean and variance or standard deviation of the probability density distribution for each dimension and the mean and variance or standard deviation of the reference probability density distribution. [Explanation of Symbols]

[0063] 14…PC for data storage and analysis 142, 142a... Anomaly detection processing unit 1421...Data acquisition unit 1422...Frequency Analysis Unit 1423, 1423a... Interval extraction section 1424...Feature calculation unit 1425, 1425a... Probability density distribution calculation unit 1426…Abnormality calculation unit 1427…Abnormality determination unit

Claims

1. A frequency analysis unit performs frequency analysis on a time series of motor current values, with each period defined as one cycle, for multiple cycles. A probability density distribution calculation unit that calculates a probability density distribution for each of the multiple intervals set on the frequency axis based on the frequency analysis results for the multiple periods, An anomaly calculation unit that calculates the anomaly degree based on the probability density distributions for each of the multiple intervals obtained and a reference probability density distribution, Equipped with, The probability density distribution calculation unit uses the average value or peak value multiplier of the gain for each interval obtained as a result of the frequency analysis as a random variable to calculate the probability density distribution for each of the multiple intervals. Motor current analyzer.

2. A probability density distribution calculation unit calculates a probability density distribution for each of several intervals set on the time axis of one period, based on multiple periods of a time series of motor current values, with a predetermined period being one period. An anomaly calculation unit that calculates the anomaly degree based on the probability density distributions for each of the multiple intervals obtained and a reference probability density distribution, Equipped with, The probability density distribution calculation unit calculates the probability density distribution for each of the multiple intervals, using the time series of the motor current value as a random variable. Motor current analyzer.

3. The time series of the motor current values ​​mentioned above was measured under multiple different load conditions. The degree of abnormality is calculated for each load condition, An abnormality determination unit estimates the nature of an abnormality by comparing the multiple abnormality levels for each of the aforementioned intervals and for each of the multiple load states with a predetermined abnormality determination value. The motor current analyzer according to claim 1 or claim 2, further comprising:

4. The anomaly calculation unit calculates the anomaly score based on the mean and variance or standard deviation of a one-dimensional probability density distribution obtained by dimensionality reduction of the multidimensional probability density distribution obtained by treating the probability density distributions for each of the multiple intervals as a multidimensional probability density distribution, and the mean and variance or standard deviation of the reference probability density distribution. The motor current analyzer according to claim 3.

5. The anomaly calculation unit calculates multiple anomaly scores based on the mean and variance or standard deviation of the probability density distribution for each of the multiple intervals as a multidimensional probability density distribution, and the mean and variance or standard deviation of the probability density distribution for each dimension, and the mean and variance or standard deviation of the reference probability density distribution. The motor current analyzer according to claim 4.

6. The steps include: performing frequency analysis on a time series of motor current values, where one period is defined as a predetermined period, for multiple periods; Based on the frequency analysis results for the aforementioned multiple periods, the step of determining the probability density distribution for each of the multiple intervals set on the frequency axis, The process includes the step of calculating the degree of anomaly based on the probability density distributions for each of the obtained intervals and a reference probability density distribution, In the step of determining the probability density distribution, the average value or peak value multiplier of the gain for each interval obtained as a result of the frequency analysis is used as a random variable to determine the probability density distribution for each of the multiple intervals. Methods for analyzing motor current.

7. The steps include: determining the probability density distribution for each of several intervals set on the time axis of one period, based on multiple periods of the time series of motor current values, with each period being a predetermined period; The process includes the step of calculating the degree of anomaly based on the probability density distributions for each of the obtained intervals and a reference probability density distribution, In the step of determining the probability density distribution, the time series of the motor current value is used as a random variable, and the probability density distribution is determined for each of the multiple intervals. Methods for analyzing motor current.

8. The steps include: performing frequency analysis on a time series of motor current values, where one period is defined as a predetermined period, for multiple periods; Based on the frequency analysis results for the aforementioned multiple periods, the step of determining the probability density distribution for each of the multiple intervals set on the frequency axis, A program that causes a computer to perform the steps of calculating the degree of anomaly based on the probability density distributions for each of the multiple intervals obtained and a reference probability density distribution, In the step of determining the probability density distribution, the average value or peak value multiplier of the gain for each interval obtained as a result of the frequency analysis is used as a random variable to determine the probability density distribution for each of the multiple intervals. program.

9. The steps include: determining the probability density distribution for each of several intervals set on the time axis of one period, based on multiple periods of the time series of motor current values, with each period being a predetermined period; A program that causes a computer to perform the steps of calculating the degree of anomaly based on the probability density distributions for each of the multiple intervals obtained and a reference probability density distribution, In the step of determining the probability density distribution, the time series of the motor current value is used as a random variable, and the probability density distribution is determined for each of the multiple intervals. program.

Images