DIAGNOSTIC DEVICE AND DIAGNOSTIC PROCEDURES

Abstract

Diagnostic device for diagnosing a condition of a diagnostic target powered by a three-phase alternating current source, the diagnostic device comprising the following: a recording unit that simultaneously records an instantaneous value of physical data in several phases of the three-phase alternating current; a conversion unit that discretizes the instantaneous value captured by the acquisition unit; an integration unit that obtains an occurrence density of a combination for each combination of discrete values ​​of the instantaneous values ​​of the physical data of the corresponding phases from a discrete value obtained by discretizing the conversion unit; an accumulation unit that collects occurrence density data, which accumulates the occurrence density for each combination obtained by the integration unit; and a diagnostic unit that diagnoses a condition of the diagnostic target based on the occurrence density data accumulated in the accumulation unit.

Description

Technical field

[0001] The present invention relates to a technology for diagnosing devices and equipment that operate on a three-phase alternating current power source. Technical background

[0002] If a rotating machine, such as a motor (electric motor) or generator installed in a production plant, suddenly fails, unplanned repairs or replacements of the lathe are required, necessitating a reduction in the plant's operating rate and a review of the production schedule. Similarly, if a power converter or cable connected to the lathe fails, unplanned repairs or replacements are also required, the plant's operating rate is reduced, and the production schedule must be reviewed.

[0003] To prevent catastrophic failures of the lathe system (the lathe and its auxiliary equipment (cables, power converters)), the lathe system is appropriately stopped and diagnosed offline to check the degree of wear, thus partially preventing a catastrophic failure. However, since the diagnosis is performed offline, it is necessary to stop the lathe system, which reduces the operating rate of the production plant. Additionally, depending on the type of wear, some wear may only become apparent when a voltage is applied. Therefore, there is a need to diagnose the condition of the lathe based on information about the current state of the lathe system.

[0004] Motor Current Signature Analysis (MCSA) is a technique for diagnosing equipment and systems based on current information from a lathe system. According to MCSA, specific frequency components in the current's frequency spectrum are detected, corresponding to factors such as rotor bar damage, rotor eccentricity, stator core damage, winding short circuits, bearing wear, etc., in order to identify faults or wear.

[0005] Additionally, PTL 1 discloses a method in which vibration sensors are installed at two locations, wherein, in particular during bearing diagnostics, vibration sensor data are acquired from these vibration sensors and an anomaly is determined from a temporary change in a path inclination or radius of a Lissajous figure, which shows that the instantaneous value of all vibration sensor data on each axis is graphically represented. List of prior art patent literature

[0006] PTL 1: JP 2000 - 258 305 A Summary of the invention: Technical problem

[0007] However, the MCSA and the technology revealed in PTL 1 have the following problems.

[0008] The MCSA must accurately detect a specific frequency component. To achieve this, it is necessary to measure the current value at a high sampling rate over an extended period. Expensive data logging equipment is required to perform long-term measurements at high sampling rates, which increases diagnostic costs.

[0009] Additionally, there is a motor fault where the phenomenon only occurs for a short time. To detect this phenomenon without faults, it is necessary to measure the current at a high sampling rate for an extended period and accumulate data. Therefore, the amount of data to be stored becomes large, and a storage device for storing the data requires a high-capacity storage unit, which increases costs. Furthermore, if the data volume becomes large, it can consume a significant amount of bandwidth on the communication path and hinder communication with other devices when the data is transferred to the cloud or a local device.

[0010] Additionally, in the technique disclosed in PTL 1, since the motor fault diagnosis is based on data obtained from a vibration sensor, it is necessary to install the vibration sensor in a position where a change in vibration is sensitive to motor faults, e.g., a position above the motor, and the installation position of the vibration sensor is limited. Furthermore, the cost of the diagnosis is increased because of the vibration sensor and the expensive data recording device, which has a sampling rate sufficient to obtain the path of the Lissajous figure.

[0011] One object of the invention is to create a technique to keep a sampling rate and data volume of current measurements low and to effectively diagnose a diagnostic target powered by three-phase alternating current. Solution to the problem

[0012] A diagnostic device according to one aspect of the invention is a diagnostic device that diagnoses the state of a diagnostic target operating on a three-phase alternating current power source and comprises a sensing unit that simultaneously acquires an instantaneous value of physical data in several phases of a three-phase alternating current, a conversion unit that discretizes the instantaneous value acquired by the sensing unit, an integration unit that obtains from the discrete value obtained by the discretization of the conversion unit the occurrence density of the combination for each combination of the discrete values ​​of the instantaneous values ​​of the physical data of each phase, an accumulation unit that stores the occurrence density data, which is the occurrence density for each combination obtained by the integration unit, and a diagnostic unit that determines the state of the diagnostic target based on the occurrence density data.diagnosed the substances that had accumulated in the accumulation unit. Advantageous effects of the invention

[0013] According to one aspect of the invention, it is possible to keep the sampling rate and the amount of data of the current measurement low and to effectively diagnose the condition of a diagnostic target that operates with three-phase alternating current. Brief description of the drawings [ Fig. 1] Fig. Figure 1 is a block diagram of a comparative example of a diagnostic device. [ Fig. 2] Fig. Figure 2 is a block diagram of a diagnostic device according to a first embodiment. [ Fig. 3] Fig. 3 is a diagram representing a data set that is stored in an accumulation unit and discretized in a conversion unit, which is in Fig. 2 is shown, which has been accumulated. [ Fig. 4] Fig. Figure 4 is a diagram that represents a current waveform when a spectrum appears at a frequency that is 1 Hz away from a fundamental frequency of 50 Hz of a U-phase current. [ Fig. 5] Fig. 5 is a diagram representing a wavy waveform, which is determined by the current waveform that is in Fig. 4 is shown, is generated. [ Fig. 6] Fig. Figure 6 is a diagram that represents a U-phase current waveform in a normal state. [ Fig. 7] Fig. Figure 7 is a diagram representing a W-phase current waveform under normal conditions. [ Fig. 8] Fig. Figure 8 is a diagram that shows a U-phase current waveform in a case where a sideband is generated. [ Fig. 9] Fig. Figure 9 is a diagram that shows a W-phase current waveform in a case where a sideband is generated. [ Fig. 10] Fig. Figure 10 is a diagram that shows the result of analyzing the degree of anomaly of a matrix A and a matrix B based on the current waveforms that are in Fig. 6 to Fig. 9 are shown, in relation to a matrix C, which is defined as the normal state. [ Fig. 11] Fig. Figure 11 is a block diagram of a diagnostic device according to a second embodiment. [ Fig. 12] Fig. Figure 12 is a diagram representing the degree of anomaly of a matrix A obtained from a current waveform of a lathe in normal state, obtained by a control instruction A. [ Fig. 13] Fig. Figure 13 is a block diagram of a diagnostic device according to a third embodiment. [ Fig. 14] Fig. Figure 14 is a block diagram of a diagnostic device according to a fourth embodiment. [ Fig. 15] Fig. Figure 15 is a block diagram of a diagnostic device according to a fifth embodiment. Description of the embodiments

[0014] In the following, embodiments of the present invention are described with reference to the drawings and using several examples. It should also be noted that the following embodiments are merely examples for illustrative purposes and that the present invention is not limited to the following embodiments.

[0015] When a lathe malfunctions, such as in the case of an electric motor (engine) or generator and a lathe system (which includes a cable and power converter attached to the lathe), the location and cause of the fault are manifold. For example, insulation and bearing wear, short circuits, open circuits, flooding, and similar issues are all possible causes. Additionally, the electric motor is often installed in harsh environments for extended periods, necessitating diagnostic equipment suited to these installation conditions.

[0016] Fig. Figure 1 is a diagram that presents a comparative example of a diagnostic device. In this comparative example, as shown in Fig. Figure 1 shows the diagnosis of the condition of a lathe 3, which is connected to a power source 1 via a cable 2. In the diagnostic device of this comparative example, two-phase current data is acquired in a current measuring unit 4 using current sensors 10a and 10b, which are attached to cable 2. The acquired current data is accumulated in an accumulation unit 8 as a value of a specific frequency spectrum obtained by Fourier transformation. Then, a diagnostic unit 9 diagnoses the condition of the lathe 3 based on the current data accumulated in the accumulation unit 8.

[0017] In the diagnostic device configured as described above, since the change in the specific frequency spectrum is measured using Fourier transforms, continuous measurement at a constant sampling rate is necessary. Therefore, it is required to increase the storage capacity for temporarily storing the measurement data or to increase the communication speed with the device for data storage, which necessitates an expensive device.

[0018] Additionally, since a phenomenon exists that only appears for a short period due to the error in lathe 3, if the measurement is performed for a long time at a high sampling rate to avoid missing the condition caused by the phenomenon, the amount of accumulated data will be large, requiring an expensive data storage device. Furthermore, if the amount of accumulated data is large, an expensive communication device supporting high-capacity communication will be needed to transfer the data to a cloud or a local device.

[0019] The inventors have taken into account that the intermittent sensor values ​​for two phases of the lathe's load current value are discretized with the measurement bit count for each phase, and that a diagnosis is performed by combining the instantaneous value of each phase with its occurrence density.

[0020] The current sensor data obtained from the three-phase motor, which is part of a three-phase lathe, exhibit a 120-degree deviation from each other. Therefore, when the two-phase current sensor data are combined, the Lissajous figure takes on a skewed elliptical shape. The inventors observed that when the two-phase current consists of ideally perfect continuous sinusoidal data, the elliptical shapes in the first and second cycles completely overlap. The measured waveform deviates from the ideal sinusoidal oscillation, and the sampling interval is finite. Consequently, the probability of measurement points exhibiting exactly the same phase is low.By superimposing data from multiple cycles, intermittently acquired over a predetermined time, and accumulating the combination of instantaneous values ​​and their occurrence densities, data can be accumulated in a format with a reduced data volume compared to accumulating two-phase waveforms as time series data. Furthermore, by superimposing data from multiple cycles, even if the sampling interval is not constant or the sampling rate is slow, the combination of instantaneous values ​​and their occurrence densities can be consistent, provided the operating conditions of the diagnostic target remain unchanged. This eliminates the need for equipment such as expensive data recording and storage systems. The crucial point here is the time synchronization of the first and second phases; these phases must necessarily be simultaneous or occur at a fixed interval, which can be arbitrarily defined.Fluctuations in the measurement interval between the first and second phases directly lead to a decrease in diagnostic accuracy. Simultaneous measurement or constant-time measurement can be achieved using a device with excellent real-time capabilities, such as a microcomputer.

[0021] The diagnostic device of this embodiment is a diagnostic device that diagnoses the state of a diagnostic target operating on a three-phase AC power source, and a sensing unit that simultaneously acquires instantaneous values ​​of physical data in several phases of a three-phase AC current, a conversion unit that discretizes the instantaneous value acquired by the sensing unit, an integration unit that obtains from the discrete value obtained by the discretization of the conversion unit the occurrence density of the combination for each combination of the discrete values ​​of the instantaneous values ​​of the physical data of each phase, an accumulation unit that stores the occurrence density data, which is the occurrence density for each combination obtained by the integration unit, and a diagnostic unit that determines the state of the diagnostic target based on the occurrence density data.which were accumulated in the accumulation unit, diagnosed, contains.

[0022] Additionally, the diagnostic device, configured as described above, can also be integrated into a lathe system. In particular, it is preferred to use a current sensor, a current measuring unit, and the like, intended for controlling the lathe, in the power converter to reduce the number of components. Furthermore, several lathes can be connected to the power converter.

[0023] Furthermore, even if the drive condition of the lathe changes by classifying and evaluating the current data for each frequency condition of the current value measured by the current sensor and the current measuring unit, it is possible to appropriately diagnose the condition of the lathe system.

[0024] More specific examples of the diagnostic device described above are described below. First embodiment

[0025] Fig. Figure 2 is a block diagram of the diagnostic device according to a first embodiment.

[0026] As in Fig. As shown in Figure 2, the diagnostic device according to this embodiment diagnoses the condition of the lathe 3 to be diagnosed, which is electrically connected to the power source 1 via cable 2 and includes the current measuring unit 4, a conversion unit 5, a measurement bit input unit 6, an integration unit 7, the accumulation unit 8, and the diagnostic unit 9. The diagnostic device can include a processor and memory, and the processor can use the memory to execute a software program that defines the operation of each of the units described above.

[0027] Power source 1 outputs a three-phase AC voltage. There are two types of three-phase AC voltage output: one where the timing of the inverter's switching element is controlled so that the motor speed and torque are the desired values, and the other where a standard power source is directly connected.

[0028] The current measuring unit 4 can be described as a sensing unit. Using current sensors 10a and 10b attached to cable 2, the current measuring unit 4 simultaneously acquires instantaneous values ​​of current data, which are physical data, in several phases of a three-phase alternating current flowing through cable 2. At this time, the current measuring unit 4 acquires the instantaneous value of the current data at any desired interval and for any desired period, while maintaining time synchronization between multiple phases.

[0029] The conversion unit 5 discretizes the instantaneous value, which was detected by the current measuring unit 4, with the set bit number A, which was set by the measurement bit number input unit 6.

[0030] The integration unit 7 has a size of 2 A (2 to the power of A) × 2 A (2 to the power of A), which corresponds to the phases of the rows and columns from the discrete values ​​obtained by discretizing conversion unit 5. By generating a matrix representing the occurrence density of the combination of the discrete values ​​of the instantaneous values ​​of the current data of each phase using each component of the rows and columns, the occurrence density of the combination is obtained for each combination of the discrete values ​​of the instantaneous values ​​of the current data of the corresponding phases.

[0031] The accumulation unit 8 collects the occurrence density data, which indicates the occurrence density for each combination obtained by the integration unit 7.

[0032] The diagnostic unit 9 diagnoses the condition of the lathe 3 and peripheral equipment such as a power converter connected to the lathe 3, based on the occurrence density data accumulated in the accumulation unit 8.

[0033] The following describes a diagnostic procedure for diagnosing the condition of lathe 3 using the diagnostic device configured as described above.

[0034] First, the current measuring unit 4 acquires current data in two phases of the three-phase alternating current flowing through cable 2 using current sensors 10a and 10b attached to cable 2. At this point, the current measuring unit 4 acquires the instantaneous value of the current data at any desired interval for any desired period, while maintaining time synchronization between the multiple phases. However, the current sensors 10a and 10b, which are being monitored by the current measuring unit 4, do not necessarily have to operate at a constant sampling interval and do not necessarily have to provide continuous readings. Furthermore, it is preferred that the interval between the acquisition of current data by current sensor 10a and the acquisition of current data by current sensor 10b be constant to allow for some fluctuation.By using a measuring device that excels in real-time processing, it is possible to establish a constant data acquisition interval that allows for a certain degree of fluctuation. Such a measuring device could, for example, utilize a microcomputer.

[0035] Furthermore, by keeping the interval between the acquisition of current data by current sensor 10a and the acquisition of current data by current sensor 10b constant, there is no need to perform continuous measurement. For example, it is possible to design the current measuring unit 4 such that a certain amount of data is accumulated in the microcomputer's memory, the stored data is saved in the storage device, and then the microcomputer's memory is cleared to restart data accumulation. A system using a generally applicable current sensor or memory can be employed.

[0036] Then, in the conversion unit 5, the instantaneous value acquired by the current measuring unit 4 is digitized with the set bit number A, which was set in the measurement bit input unit 6. The integration unit 7 receives the occurrence density of a combination for each combination of the discrete values ​​of the instantaneous values ​​of the two-phase current data from the discrete values ​​obtained by the discretization in the conversion unit 5. The occurrence density data, which indicates the occurrence density of each combination, are accumulated in the accumulation unit 8. At this point, in the conversion unit 5, the instantaneous value of the two-phase current data is discretized with a predefined bit number A, and in the integration unit 7, the rows and columns have a phase 2. A (2 to the power of A) × 2 A(2 to the power of A), corresponding to each phase. If the occurrence density is obtained by generating a matrix that specifies the occurrence density of the combination of the discrete values ​​of the instantaneous values ​​of the current data for each phase using each component of the rows and columns, the amount of current data for each phase can be reduced to a size of 2 A × 2 A be compressed.

[0037] Here, the description of the data set through discretization is given in a case where the control pattern for lathe 3 does not change, lathe 3 operates at a constant fundamental frequency, and an ideal sinusoidal oscillation current flows through lathe 3. Furthermore, the measurement bit count of the data recording device is 8 bits.

[0038] Fig. Figure 3 is a diagram representing the amount of data that is discretized in the accumulation unit 8 by the conversion unit 5, which is in Fig. 2 is shown, which has been accumulated.

[0039] The amount of data is then set to 1 if the two-phase current is measured for 200 seconds at a sampling rate of 200 Hz.

[0040] As a first step, a waveform, serving as current data acquired by the current measuring unit 4, is discretized with the number of bits 8, which is set by the measurement bit input unit 6 in the conversion unit 5. The integration unit 7 then receives the occurrence density of current values, discretized with 8 bits, i.e., from 1 to 257 for each matrix element of the 28 × 28 matrix. For example, in a measurement time of 200 seconds, the value obtained by dividing the number of times the discretized current value of current sensor 10a (257) and the discretized current value of current sensor 10b (257) by 40,000 measurement points is given. The 200-second measurement time need not be continuous, and if the control pattern for the lathe 3 does not change and the lathe 3 operates at a constant fundamental frequency, the measurement can be performed in such a way that it is divided into several segments.As a result, even a low-speed measuring device with a small amount of storage can capture data needed for diagnosis.

[0041] In this case, the data volume is 0.81, and a 19% reduction in data volume can be achieved. As a result, even a low-speed measuring device with a small amount of memory can acquire the data necessary for diagnostics. In this embodiment, the value obtained by dividing the number of measurement points in each matrix element is stored. However, the number of occurrences is also stored, and the number of measurement points can be stored separately in a format linked to a file containing the number of occurrences. Furthermore, if the measurements are divided into multiple iterations, the accumulation of errors due to integration can be reduced by storing the number of occurrences and the number of measurement points separately.

[0042] Furthermore, as a second method, instead of preparing a 28 × 28 matrix in advance, the data is stored in the form (discreteized current value 1 of current sensor 10a, discretized current value 1 of current sensor 10a, number of occurrences), (discreteized current value 1 of current sensor 10a, discretized current value 2 of current sensor 10a, number of occurrences), etc. In the second method, the data ultimately obtained consists of data that excludes the components of the matrix elements from the first method where the values ​​of the matrix elements are 0 and there is no need to prepare an empty cell. Therefore, only the value of the component in which a non-zero numerical value is stored is accumulated in the accumulation unit 8, and the amount of accumulated data can be further reduced. It should also be noted that the time required to exclude the zero component is not particularly limited.

[0043] As a data accumulation method, the data can be accumulated in the form (discreteized current value 1 of current sensor 10a, discretized current value 1 of current sensor 10a, number of occurrences), (discreteized current value 1 of current sensor 10a, discretized current value 2 of current sensor 10a, number of occurrences), etc. Once the matrix of the first method is established, the data can be transformed into a form (discreteized current value 1 of current sensor 10a, discretized current value 1 of current sensor 10a, number of occurrences), (discreteized current value 1 of current sensor 10a, discretized current value 2 of current sensor 10a, number of occurrences), etc.

[0044] The diagnostic unit 9 then diagnoses the condition of the lathe 3 based on the occurrence density data accumulated in the accumulation unit 8.

[0045] Here, a description of a diagnostic procedure for the diagnostic unit 9 is given in a case where a spectrum is generated at a certain frequency for some reason when the control pattern for the lathe 3 does not change and the lathe 3 operates at a constant fundamental frequency.

[0046] Fig. Figure 4 is a diagram representing a current waveform when a spectrum appears at a frequency that is 1 Hz away from a fundamental frequency of 50 Hz of a U-phase current, and Fig. 5 is a diagram representing a wavy waveform, which is determined by the current waveform that is in Fig. 4 is shown, is generated.

[0047] For example, when a sideband wave of the fundamental frequency is generated due to deterioration of bearing wear, causing a spectrum to appear, for some reason, at a frequency 1 Hz away from the fundamental frequency of 50 Hz of the U-phase current, as in Fig. As shown in Figure 4, the U-phase current appears as a waveform exhibiting a rise of 1 Hz, as in Fig. 5 is shown.

[0048] In the diagnostic device, which is in Fig. As shown in Figure 1, to accurately separate the 50 Hz and 51 Hz components from this waveform when the current flowing through cable 2 is measured at a frequency of 200 Hz, it is necessary to continuously measure 20,000 data points for at least 200 seconds. Furthermore, the frequency of sideband vibrations increases with wear and is low in the early stages of wear, necessitating long-term data measurement. Therefore, a special and expensive measuring device is required to perform this high-precision separation.

[0049] On the other hand, the diagnostic device, which is in Fig. As shown in Figure 2, the current measuring unit 4 records the current values ​​of two or more phases at any intervals for any period of time, while maintaining time synchronization between the phases. The conversion unit 5 discretizes the current data of two or more phases for each phase using the number of bits set in the measurement bit input unit 6, and the occurrence count or occurrence density of the discretized current data is added to each matrix element in the integration unit 7. Subsequently, the diagnostic unit 9 compares the data, which has been accumulated as occurrence density data in the accumulation unit 8, and the condition of the peripheral device, such as the lathe 3 or the power converter connected to the lathe 3, is diagnosed based on the change, such that it is possible to diagnose the lathe 3 without increasing the sampling rate or the amount of data.

[0050] Additionally, in order to capture high-frequency information, it is desirable to perform a measurement asynchronously to the switching time sequence of the power converter.

[0051] Furthermore, when the sampling rate is reduced, it is desirable for the current measuring unit 4 to measure the current flowing through cable 2 asynchronously to the fundamental frequency using the current sensors 10a and 10b. Specifically, at a sampling rate with a period that is an integer multiple of the fundamental frequency, only a value of a specific phase is obtained. Therefore, in the case of wear where a change occurs only in a specific phase, there is a risk of missing data. For this reason, it is desirable for the sampling rate to be a frequency that is different from an integer multiple of the period of the fundamental frequency. As a result, it is possible to obtain the same matrix as if the data were obtained at a high sampling rate.

[0052] Fig. Figure 6 is a diagram representing a U-phase current waveform in a normal state, and Fig. Figure 7 is a diagram representing a W-phase current waveform under normal conditions. Additionally, Fig. 8 a diagram that represents a U-phase current waveform in a case where a sideband is generated, and is Fig. Figure 9 shows a diagram representing a W-phase current waveform in a case where a sideband is generated. Furthermore, the sampling rate is 200 Hz and the data for 1 second out of 100 seconds of data acquisition time are magnified.

[0053] In a 50 Hz sine wave under normal conditions without wave motion, the U-phase current waveform is as shown in Fig. Figure 6 shows the W-phase current waveform as in Fig. Figure 7 shows the U-phase and W-phase currents being phase-shifted by 120 degrees relative to each other.

[0054] If the waveforms that are in Fig. 6, Fig. 7, Fig. 8 and Fig. 9. The current waveform measured by the current measuring unit 4 is discretized with the number of bits 8 set in the measurement bit input unit 6, as described in the first procedure, and the integration unit 7 stores the occurrence densities of the current values ​​discretized with 8 bits, i.e., 1 to 257 for each matrix element of the matrix, which has a size of 28 × 28, in matrix A and matrix B, respectively. Furthermore, matrix A uses the waveforms that are in Fig. 6 and Fig. Matrix B, shown in Figure 7, serves as the basis and corresponds to the matrix in the normal state. Matrix B uses the waveforms shown in Figure 7. Fig. 8 and Fig. The figures shown in 9 serve as a basis and represent the matrix in its deteriorated state.

[0055] Diagnostic unit 9 compares matrix A and matrix B to matrix C, which was predefined as the normal state through learning, and calculates how close matrix A and matrix B are to matrix C in order to determine any deviation from the normal state, i.e., the degree of anomaly. As wear progresses, the fluctuation of the current value increases, and consequently, the frequency of occurrence in the area with a high frequency decreases, while the current value is distributed over the area with a low frequency. As a result, the value of each matrix element changes, the number of matrix elements with a value of 0 decreases, and the matrix deviates from the matrix defined as the normal state.

[0056] After performing the diagnosis described above, diagnostic unit 9 outputs a diagnostic result. A suitable method for sending the diagnostic result to the user can be selected. Examples of methods for sending the result to the user include illuminating a lamp, sending an email notification, and the like, in addition to displaying the result on the display device. Other possible methods include (1) a method for displaying the matrix to be diagnosed on the screen and allowing the user to determine whether a correspondence exists, (2) a method for quantifying the difference between the matrices to be diagnosed and communicating this difference to the user, and (3) a method for notifying the user if the threshold is exceeded.

[0057] Here, as a method for digitizing the difference between the matrices to be diagnosed, the application of machine learning can be considered in point (2) above. It may be necessary to choose a machine learning algorithm that clearly identifies the difference between the diagnostic target matrices. For example, if the Bhattacharyya coefficient (Bhattacharyya distance) Z of matrix A, which is to be diagnosed with respect to matrix C, defined as the normal state, is defined by the following equation, assuming the matrix elements of matrix AA i,j are and the matrix elements of matrix CC i,j are. Z=∑i,jAi,jCi,j

[0058] In the equation given above, if the Bhattacharyya coefficient Z is large, it can be diagnosed that the condition deviates from the normal state, i.e., that wear is progressing. If the coefficient Z is small, it can be diagnosed that the condition is normal. It can be stated that the Bhattacharyya coefficient Z indicates the degree of anomaly. As described above, the matrix data is stored in the accumulation unit 8, and the diagnostic unit 9 compares the matrix stored in the accumulation unit with the reference matrix, which is defined as the normal state, to determine the normality of the condition. This makes it easy to know how much the condition deviates from the normal state and to diagnose normality.

[0059] When using a matrix where the number of occurrences is stored in each matrix element, it is necessary to compare the normal state and the measurement points of the matrix to be diagnosed. A local subspace method can be used as a diagnostic procedure in this case. In the local subspace method, for each matrix element of the matrix to be diagnosed, two nearest points are chosen from the two-dimensional space defined by the rows and columns of the matrix, which is defined as the normal state. The degree of wear is then defined by the distance between the line connecting these two points and the point to be diagnosed. In some cases, it is effective to weight the occurrence density of normal values.Particularly in the case of data that can be identified as an anomalous state, it is desirable that the discriminant can be adjusted in such a way that the matrix between the normal state and the anomalous state can be easily distinguished.

[0060] Additionally, in a case where the local subspace method is used, besides a method for calculating distances for all matrix elements to be diagnosed and quantifying a change in the matrix, any evaluation method based on a fluctuation error of a current waveform can be chosen, such as quantifying an average value of all distances of all points of the diagnostic target or a numerical value expressed by the distance of only a specific phase point. Additionally, in a case where computational speed is a priority, grouping methods such as vector quantization grouping and K-means grouping can be used. Furthermore, a method known as a deep neural network, which is a method for automatically searching for a feature based on a large dataset, can be applied.

[0061] Fig. Figure 10 is a diagram that shows the result of analyzing the degree of anomaly of a matrix A and a matrix B based on the current waveforms that are in Fig. 6, Fig. 7, Fig. 8 and Fig. 9 are shown, in relation to a matrix C, which is defined as the normal state.

[0062] Matrix A based on the current waveform, which Fig. 6 and Fig. 7 is shown, and matrix B is based on the current waveform shown in Fig. 8 and Fig. Figure 9 shows that the degree of anomaly can be analyzed from the Bhattacharyya distance Z according to the equation described above. As shown in Fig. As shown in Figure 10, the Bhattacharyya distance Z of matrix A does not exceed a preset threshold, such that the diagnostic unit 9 determines that the Bhattacharyya distance is normal. On the other hand, the Bhattacharyya distance Z of matrix B exceeds the preset threshold, as shown in Figure 10. Fig. 10 is shown. Thus, diagnostic unit 9 determines that it is anomalous.

[0063] As described above, depending on whether the Bhattacharyya distance Z exceeds a preset threshold, it is possible to detect an increase in the degree of anomaly before lathe 3 fails.

[0064] As described above, since the occurrence densities of the combinations of instantaneous values ​​of the current data for each phase are integrated and accumulated, the amount of data to be saved can be kept low. Additionally, since the condition of lathe 3, which is to be diagnosed, is diagnosed based on the integrated data obtained by integrating the occurrence densities of the combinations of instantaneous values ​​of the current data for each phase, it is not necessary to measure the current data at a high sampling rate.

[0065] Furthermore, this embodiment describes a case in which a spectrum is generated at a specific frequency. However, even if the type of wear does not generate a spectrum at a specific frequency, the wear can cause a change in the load of the lathe 3 or a change in impedance to alter the current. It is possible to detect an anomaly using the diagnostic device located in Fig. Figure 2 is shown, and the diagnostic procedure described above is used to detect it. Deterioration that does not appear as a change in the spectrum of a specific frequency is specifically defined as deterioration other than deterioration that appears as a peak value of a specific frequency detectable by MCSA, i.e., grease deterioration, thermal deterioration of insulation material, and moisture absorption.

[0066] In addition to the lathe 3, the diagnostic device of this embodiment also allows for the diagnosis of a motor system that includes a device electrically or mechanically connected to the lathe 3, such as a cable, a power converter, and a load. In a motor system containing peripheral devices connected to the lathe 3, even if a peripheral device other than the lathe 3 fails or wears out, the current flowing through the lathe 3 changes due to alterations in the impedance and load of these devices, such that the wear can be detected by this method. Second embodiment

[0067] Next, a case is described in which the control pattern for lathe 3 changes. In a case where the control pattern for lathe 3 changes, the fundamental frequency, the carrier frequency, and the like of lathe 3 change such that the diagnostic device detects the change as an anomaly that is in Fig. As shown in Figure 1, a diagnosis can be made. Additionally, if learning is performed, since the normal state includes a condition in which the control pattern is modified, a change due to wear and tear may be overlooked, and an incorrect report may be generated. Therefore, it is desirable to diagnose the normality for each control pattern.

[0068] Fig. Figure 11 is a block diagram of a diagnostic device according to the second embodiment.

[0069] As in Fig. As shown in 11, the diagnostic device in this embodiment differs from the diagnostic device shown in 11. Fig. Figure 2 differs in that an instruction unit 13 is provided.

[0070] The instruction unit 13 sends a control instruction, specifying information about the control pattern, to the integration unit 7. The control instruction can be any instruction value that can be output by the power source 1, and comparable values ​​such as a voltage instruction value, a current instruction value, an excitation current instruction value, a torque current instruction value, a speed instruction value, a frequency instruction value, and the like.

[0071] The integration unit 7 receives the occurrence density, one combination for each combination for each control instruction to the lathe 3, and the accumulation unit 8 collects occurrence density data for each control instruction.

[0072] The following describes a case in which the state of lathe 3 is diagnosed under two conditions: control instruction A and control instruction B, where the voltage instruction value and the frequency instruction value differ. Furthermore, control instruction A corresponds to a fundamental current frequency of 50 Hz, and control instruction B corresponds to a fundamental current frequency of 100 Hz.

[0073] The integration unit 7 integrates the occurrence density obtained by the conversion unit 5 for each current value distributed by control instruction A and control instruction B based on the fundamental frequency input to instruction unit 13. Then, after data for a randomly determined period or score has been accumulated, the data are stored in the accumulation unit 8 as the occurrence density data.

[0074] The diagnostic unit 9 then compares the occurrence density data accumulated in the accumulation unit 8 with data defined as normal, obtained using the same control instruction information, to determine the degree of anomaly of the target data and diagnoses whether the lathe 3 is normal.

[0075] Fig. Figure 12 is a diagram representing the degree of anomaly of matrix A, which is obtained from a current waveform of a lathe in its normal state, obtained by control instruction A. Furthermore, matrices α and β are matrices defined by current waveforms of lathe 3 in its normal state, measured by control instructions A and B, respectively.

[0076] As in Fig. As shown in Figure 12, when the lathe 3 is diagnosed in its normal state using matrix α with matrix A based on the current data of the lathe 3 in its normal state, the Bhattacharyya distance Z of matrix A does not exceed a predetermined threshold, which the diagnostic unit 9 determines to be normal.

[0077] On the other hand, if the lathe 3 is detected in the normal state using matrix β with matrix A based on the current data of the lathe 3 in the normal state, the Bhattacharyya distance Z of matrix A exceeds the specified threshold, as in Fig. 12 is shown. Therefore, diagnostic unit 9 determines this to be anomalous, even though lathe 3 is in its normal state.

[0078] As described above, in this embodiment, to diagnose the state of the lathe 3 based on the matrix composed of the U-phase and W-phase currents classified as the state of the same control instruction, the control pattern and current information are combined to improve diagnostic accuracy by classifying the current waveform based on the instruction input into the instruction unit 13 and integrating the occurrence density for each classified data element. This allows for a suitable diagnosis to be performed for each control state.

[0079] Furthermore, in integration unit 7, it is not always necessary to use all instruction values ​​that can be output by power source 1; only those instruction values ​​that are highly sensitive to the wear of the detection target can be used. The highly sensitive instruction value can be selected by comparing the degree of similarity between the matrix in the deteriorated state and the matrix in the normal state, such that the difference between the deteriorated and normal states is easily discernible. Additionally, if the number of conditions is increased, the amount of data will increase by the amount classified by the conditions, even if the current data has the same measurement time. Therefore, the number of instruction values ​​to be used can be limited by balancing diagnostic accuracy with the amount of data. Third embodiment

[0080] Then, if the information from instruction unit 13 is not as in the diagnostic device that is in Fig. 11, which can be designated, the characteristic frequency of the current waveform of the lathe 3 can be used instead.

[0081] Fig. Figure 13 is a block diagram of the diagnostic device according to the third embodiment.

[0082] As in Fig. As shown in 13, the diagnostic device of this embodiment differs from the diagnostic device shown in 13. Fig. Figure 2 differs in that a frequency extraction unit 12 is provided.

[0083] Then, when the control pattern of the lathe 3 is recognized based on the characteristic frequency of the current waveform of the lathe 3, the output of the current measuring unit 4 is branched and fed into the frequency extraction unit 12, as shown in Fig. Figure 13 illustrates this. The frequency extraction unit 12 extracts a characteristic frequency, such as the fundamental frequency and / or the carrier frequency. The integration unit 7 then obtains an occurrence density of the combination for each combination under each measurement condition that exhibits the same characteristic frequency. The accumulation unit 8 collects the occurrence density data, which are stored for each characteristic frequency.

[0084] As described above, in this embodiment, to diagnose the state of the lathe 3 based on the matrix composed of the U-phase and W-phase currents classified as the state of the same control instruction, the output of the current measuring unit 4 is branched to extract the characteristic frequency using the frequency extraction unit 12. The current waveform is then classified based on the characteristic frequency, and the occurrence density is accumulated for all classified data elements. This allows the control pattern and current information to be combined to improve diagnostic accuracy, enabling a suitable diagnosis for each control state. Fourth embodiment

[0085] Fig. Figure 14 is a block diagram of the diagnostic device according to the fourth embodiment.

[0086] As in Fig. As shown in 14, the diagnostic device in this embodiment differs from the one shown in 14. Fig. Figure 2 differs in that several lathes 3-1 to 3-n are connected to a power source 1.

[0087] In this embodiment, the diagnosis is performed using the measurement results of the load currents of all lathes 3-1 to 3-n by the current sensors 10a and 10b, which are attached to cable 2 connecting the power source 1 and the lathes 3-1 to 3-n. The diagnostic procedure is the same as when only one lathe is connected. In addition to classification using the control instruction value and reflection of the instruction value information, the diagnosis is performed using the measured diagnostic data as the distribution of Lissajous figures.

[0088] Furthermore, since the diagnosis is performed using the current data of the multiple lathes 3-1 to 3-n as a single load current, the change in the current waveform of the deteriorating lathe is reduced by the current waveforms of the other lathes. Therefore, it is desirable to evaluate the degree of anomaly using machine learning, in particular the local subspace method. Fifth embodiment

[0089] Fig. Figure 15 is a block diagram of a diagnostic device according to a fifth embodiment.

[0090] As in Fig. As shown in 15, the diagnostic device in this embodiment differs from the one shown in 15. Fig. Figure 2 differs in that each three-phase load current supplied from the power source 1 to the lathe 3 is measured by the current sensors 10a to 10c.

[0091] In this embodiment, the current measuring unit 4 simultaneously acquires the instantaneous value of the current data in the three phases of the three-phase alternating current, and the integration unit 7 uses the discrete value obtained by the discretization of the conversion unit 5 to obtain the occurrence density of the combination for each combination of the discrete values ​​of the instantaneous current data of the three phases. In this case, when comparing distributions, the distribution of the Lissajous figure can be obtained and evaluated using multiple combinations of two current sensors selected from three or more. Additionally, a multidimensional matrix (a tensor) can be defined using three or more current sensors, and motor wear can be assessed based on changes in the tensor.

[0092] In the case of the three-phase AC lathe 3, in addition to the U-phase, V-phase and W-phase load currents, the current values ​​measured by the current sensors 10a to 10c include a current with high diagnostic sensitivity, such as a zero-phase current measured by clamping the three phases, a two-phase current measured by clamping any two phases, a leakage current measured by clamping both the winding start and the winding end of a motor winding, and a current flowing from the motor to ground, which can be chosen arbitrarily and the current sensor can be installed at this position.

[0093] As described above, since the condition of lathe 3 is diagnosed based on the three-phase current data of the three-phase alternating current, a more accurate diagnosis can be carried out. Reference symbol list 1 power source 2 cables 3 lathe 4 Current measuring unit 5 conversion unit 6 Measurement bit input unit 7 Integration Unit 8 accumulation units 9 Diagnostic Unit 10a, 10b, 10c Current sensor 12 Frequency Extraction Unit 13 Instruction Unit

Claims

[1] Diagnostic device for diagnosing a condition of a diagnostic target powered by a three-phase alternating current source, the diagnostic device comprising: a recording unit that simultaneously records an instantaneous value of physical data in several phases of the three-phase alternating current; a conversion unit that discretizes the instantaneous value captured by the acquisition unit; an integration unit that obtains an occurrence density of a combination for each combination of discrete values ​​of the instantaneous values ​​of the physical data of the corresponding phases from a discrete value obtained by discretizing the conversion unit; an accumulation unit that collects occurrence density data, which accumulates the occurrence density for each combination obtained by the integration unit; and a diagnostic unit that diagnoses a condition of the diagnostic target based on the occurrence density data accumulated in the accumulation unit. [2] Diagnostic device according to claim 1, wherein The conversion unit discretizes the instantaneous value of the physical two-phase data with a predetermined number of bits A and the integration unit has a size of 2 A (2 to the power of A) × 2 A (2 to the power of A) has, where each row and column corresponds to a phase, and generates a matrix in which each component of a row and a column represents the occurrence density of a combination of discrete values ​​of the instantaneous values ​​of physical data of the corresponding phases. [3] Diagnostic device according to claim 2, wherein the accumulation unit holds the data of the matrix and The diagnostic unit diagnoses the normality of a state of the diagnostic target by comparing a matrix held in the accumulation unit with a reference matrix specified as a normal state. [4] Diagnostic device according to claim 3, wherein the diagnostic unit diagnoses a normality of a state of the diagnostic target on the basis of a difference between the matrix accumulated in the accumulation unit and the reference matrix. [5] Diagnostic device according to claim 2, wherein the accumulation unit only accumulates values ​​of components in the matrix in which non-zero numerical values ​​are stored. [6] Diagnostic device according to claim 1, wherein the diagnostic target includes a controllable lathe, the integration unit receives an occurrence density of the combination for each combination for each control instruction of the lathe and The accumulation unit collects the occurrence density data for each control instruction. [7] Diagnostic device according to claim 1, wherein the diagnostic target includes a controllable lathe, The integration unit captures an occurrence density of the combination for each combination for each characteristic frequency, which is changed by control at the lathe, and The accumulation unit collects the occurrence density data for each characteristic frequency. [8] Diagnostic device according to claim 7, wherein the characteristic frequency is a fundamental frequency and / or a carrier frequency of a current that is applied to each phase of the lathe. [9] Diagnostic device according to claim 1, wherein the acquisition unit simultaneously captures an instantaneous value of physical data in the three phases of the three-phase alternating current and The integration unit obtains an occurrence density of a combination for each combination of discrete values ​​of instantaneous values ​​of the physical three-phase data from discrete values ​​obtained by discretization of the conversion unit. [10] Diagnostic procedure for diagnosing a condition of a diagnostic target operating on a three-phase alternating current source, the diagnostic procedure comprising: Acquiring an instantaneous value of physical data in multiple phases of three-phase alternating current simultaneously; Discretizing the recorded instantaneous value; Obtaining an occurrence density of a combination for each combination of discrete values ​​of the instantaneous values ​​of the physical data of the corresponding phases from the discrete values ​​through discretization; Accumulating occurrence density data that indicates the occurrence density obtained for each combination; and Diagnosing a condition of the diagnostic target based on the accumulated occurrence density data.

Images