Abnormality diagnosis device, abnormality diagnosis system, and storage medium

Abstract

This abnormality diagnostic device for diagnosing an abnormality that occurs in a factory acquires data to be used for detecting an abnormality that occurs in a factory, detects an abnormality of a subject of diagnosis by using the data to be used for detecting an abnormality, reports the detected abnormality, acquires data to be used for analyzing an abnormality, and analyzes a candidate abnormality cause by using the data to be used for analyzing an abnormality in a period including a time point at which the abnormality has occurred, or by using the data to be used for analyzing an abnormality and the data to be used for detecting an abnormality in a period including a time point at which the abnormality has been detected.

Description

[Technical Field]

[0001] The present invention relates to an abnormality diagnosis device, an abnormality diagnosis system, and a storage medium. [Background technology]

[0002] On a factory production line, even a single malfunction can affect the entire process. To maintain a high operating rate, it is necessary to realize a system for detecting signs of malfunction that can prevent abnormalities and problems in factory equipment before they occur. In case of an accident, the production line but If a system stops due to some kind of problem, it is necessary to identify the cause as quickly as possible and resolve the problem promptly.

[0003] A conventional technique for identifying the cause of an abnormality occurring in a machine involves acquiring a sensor signal related to the physical state of the machine, determining the operating status of the machine based on the information acquired from the machine, calculating the degree of abnormality of the sensor signal for each operating status of the machine, and diagnosing the cause of the machine abnormality from historical data that is a series of the degree of abnormality for each operating status (see, for example, Patent Document 1).

[0004] The anomaly cause identification device described in Patent Document 1 detects signs of anomalies based on signals from sensors, isolates signals that can identify anomalies, and identifies the cause of the anomaly. The results of the identification are stored in memory, and the location and type of the anomaly are diagnosed. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2019-148971 Summary of the Invention [Problem to be solved by the invention]

[0006] In anomaly diagnosis, there is a trade-off between computational load and diagnostic accuracy. Increasing diagnostic accuracy increases the computational load, and in order to reduce the computational load, it is necessary to adjust the diagnostic accuracy.

[0007] In the field of abnormality diagnosis in the manufacturing industry, there is a demand for technology that reduces the computational load while ensuring diagnostic accuracy. [Means for solving the problem]

[0008] an anomaly detection data acquisition unit that acquires data to be used for anomaly detection; an anomaly detection condition selection unit that accepts selection of the data to be used for anomaly detection and the anomaly detection conditions to be used for the anomaly detection; an anomaly detection unit that uses the data acquired by the anomaly detection data acquisition unit to detect an anomaly of a diagnosis target; an anomaly notification unit that notifies of an anomaly detected by the anomaly detection unit; an anomaly analysis data acquisition unit that acquires data to be used for anomaly analysis; and an anomaly analysis unit that analyzes candidate anomaly causes using data to be used for the anomaly analysis for a time period including the time when the anomaly occurred, or data to be used for the anomaly detection for a time period including the time when the anomaly was detected, wherein the anomaly detection condition selection unit accepts selection of whether to prioritize accuracy or speed in the anomaly detection, and determines the data to be used for the anomaly detection and the model to be used for the anomaly detection based on the selection. an anomaly detection data acquisition unit that acquires data to be used for anomaly detection; an anomaly detection condition selection unit that accepts selection of the data to be used for anomaly detection and the anomaly detection conditions to be used for the anomaly detection; an anomaly detection unit that uses the data acquired by the anomaly detection data acquisition unit to detect an anomaly of a diagnosis target; an anomaly notification unit that notifies of an anomaly detected by the anomaly detection unit; an anomaly analysis data acquisition unit that acquires data to be used for anomaly analysis; and an anomaly analysis unit that analyzes candidate anomaly causes using data to be used for the anomaly analysis for a time period including the time when the anomaly occurred, or data to be used for the anomaly detection for a time period including the time when the anomaly was detected, wherein the anomaly detection condition selection unit accepts selection of whether to prioritize accuracy or speed in the anomaly detection, and determines the data to be used for the anomaly detection and a model to be used for the anomaly detection based on the selection. The computer-readable storage medium according to one aspect of the present disclosure may, when executed by one or more processors, The abnormality detection data acquisition unit operates as an abnormality detection data acquisition unit, an abnormality detection condition selection unit, an abnormality detection unit, an abnormality notification unit, an abnormality analysis data acquisition unit, and an abnormality analysis unit. The abnormality detection data acquisition unit acquires data used for detecting abnormalities occurring in a factory, and the abnormality detection condition selection unit selects the abnormality analysis data acquisition unit. Choosing whether to prioritize accuracy or speed in anomaly detection Accept Based on the selection, the data to be used for the anomaly detection and the Determine the model death, The abnormality detection unit Detecting an abnormality in a diagnosis target using the data used for the abnormality detection; The abnormality notification unit notifying the detected abnormality; The abnormality analysis data acquisition unit, Acquire data to be used for anomaly analysis, The anomaly analysis unit Analyzing candidates for abnormality factors using data used for the abnormality analysis for a period including the time when the abnormality occurred, or data used for the abnormality detection for a period including the time when the abnormality was detected and data used for the abnormality analysis. Order Remember. [Effects of the Invention]

[0009] According to one aspect of the present invention, it is possible to reduce the calculation load while ensuring the accuracy of diagnosis in machine abnormality diagnosis. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 2 is a hardware configuration diagram of the abnormality diagnosis device. [Figure 2] 1 is a block diagram of an abnormality diagnosis device according to a first disclosure. [Figure 3] 10 is an example of an abnormality detection condition selection screen. [Figure 4] FIG. 10 is a diagram illustrating the difference in the number of pieces of data between the accuracy-oriented and speed-oriented cases. [Figure 5] 10 is an example of an abnormality notification screen. [Figure 6] FIG. 10 is a diagram illustrating an example of extracting data for anomaly detection. [Figure 7] FIG. 10 is a diagram illustrating an example of cutting out data for abnormality analysis. [Figure 8] 10 is an example of a display screen of an analysis result. [Figure 9] 10 is an example of an abnormality analysis condition selection screen. [Figure 10] 4 is a flowchart illustrating the operation of the abnormality diagnosis device. [Figure 11] FIG. 10 is a diagram illustrating an abnormality diagnosis system according to a second disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0011] [First Disclosure] The fault diagnosis device 100 detects and diagnoses faults in a factory, machines and equipment in the factory, and products manufactured in the factory. The fault diagnosis device 100 may be implemented in a numerical control device, a programmable logic controller (PLC), a server, a personal computer, or other information processing device in the factory, or the components of the fault diagnosis device 100 may be distributed throughout the factory system as described in the second disclosure.

[0012] The hardware configuration of the abnormality diagnosis device 100 of the present disclosure will be described with reference to FIG. The abnormality diagnostic device 100 is an information processing device such as an abnormality detection device for factory equipment and machines installed inside a factory, a PC (personal computer) that monitors the state of the factory, or a numerical control device that monitors the state of a machine tool. The CPU 111 provided in the abnormality diagnosis device 100 is a processor that controls the entire abnormality diagnosis device 100. The CPU 111 reads a system program processed in the ROM 112 via the bus, and controls the entire abnormality diagnosis device 100 in accordance with the system program. The RAM 113 temporarily stores temporary calculation data, display data, various data input by the user via the input unit 71, etc.

[0013] The display unit 70 is a monitor or the like attached to the abnormality diagnostic device 100. The display unit 70 displays an operation screen, a setting screen, and the like of the abnormality diagnostic device 100.

[0014] The input unit 71 is a keyboard, a touch panel, or the like that is integrated with the display unit 70 or is separate from the display unit 70. The user operates the input unit 71 to input data to the screen displayed on the display unit 70. The display unit 70 and the input unit 71 may be mobile terminals.

[0015] The nonvolatile memory 114 is a memory that maintains its stored state even when the power to the abnormality diagnosis device 100 is turned off, for example, by being backed up by a battery (not shown). The nonvolatile memory 114 stores programs read from external devices via an interface (not shown), programs input via the input unit 71, and various data acquired from various parts of the abnormality diagnosis device 100 and sensors in the factory. The programs and various data stored in the nonvolatile memory 114 may be expanded into the RAM 113 when executed / used. In addition, various system programs are written in the ROM 112 in advance.

[0016] FIG. 2 is a block diagram of the abnormality diagnosis device 100 of the first disclosure. The abnormality diagnosis device 100 includes an abnormality detection data acquisition unit 11, an abnormality analysis data acquisition unit 12, an abnormality detection unit 13, an abnormality detection condition selection unit 14, an abnormality notification unit 15, a data extraction unit 16, an abnormality analysis unit 17, an analysis result notification unit 18, an abnormality analysis condition selection unit 19, a diagnosis history storage unit 20, and a diagnosis history presentation unit 21.

[0017] The anomaly detection data acquisition unit 11 acquires data to be used for anomaly detection. The data to be used for anomaly detection is selected by the anomaly detection condition selection unit 14, which will be described later. The data acquired by the anomaly detection data acquisition unit 11 includes operation data from control devices such as numerical control devices and PLCs, and sensor data detected by sensors installed in factories and internal sensors of industrial machines including machine tools.

[0018] The fault analysis data acquisition unit 12 acquires data to be used for fault analysis. The data to be used for fault analysis is selected by the fault analysis condition selection unit 19, which will be described later. The data acquired by the fault analysis data acquisition unit 12 includes operation data from control devices such as numerical control devices and PLCs, and sensor data detected by sensors installed in factories and internal sensors of industrial machinery including machine tools. Note that the data used for fault detection acquired by the fault detection data acquisition unit 11 is also included in the fault analysis data.

[0019] The anomaly detection unit 13 detects anomalies using the data acquired by the anomaly detection data acquisition unit 11. The data used for anomaly detection is time-series data such as motor torque, vibration, temperature, and pressure. Anomaly detection is performed in real time. Under normal conditions, the anomaly detection unit 13 determines whether an anomaly exists in the anomaly detection data or calculates the degree of anomaly. When detecting whether an anomaly exists or calculating the degree of anomaly, it is desirable to calculate feature quantities of the anomaly analysis data related to the occurrence of an anomaly. The feature quantities are calculated from the motor torque, vibration, control signal, etc. The feature quantities include the cut-out section of each data, statistics (Root Mean Square (RMS), maximum value, minimum value, standard deviation, skewness, kurtosis), or frequency domain values ​​after Fourier transform (Power Spectral Density (PSD), amplitude and phase at each frequency). The features are not predetermined, but are selected from normal and abnormal data based on suitability and versatility. Several feature candidates are selected in advance, and the suitability and versatility of the features are evaluated based on certain criteria, and the optimal feature is determined through repeated steps. Here, compatibility refers to the feature that most deviates between normal and abnormal, and also includes a combination of such features. For example, digital signals have higher versatility than analog signals in terms of noise resistance and portability. Also, from the perspective of ease of data acquisition, speed signals have higher versatility than torque command signals, and control signals have higher versatility than speed signals. The abnormality notification unit 15 may notify the operator of the feature amount of the abnormality analysis data. The anomaly detection unit 13 may calculate the degree of anomaly based on the feature amount of the anomaly analysis data. The anomaly notification unit 15 notifies the operator of the degree of anomaly of the anomaly analysis data calculated by the anomaly detection unit 13. The operator can check the degree of anomaly of multiple types of data.

[0020] The anomaly detection condition selection unit 14 accepts the selection of anomaly detection conditions from the operator. Fig. 3 shows an example of an anomaly detection condition selection screen. The anomaly detection condition selection screen allows the user to input the type of machine or equipment to be diagnosed, the parts to be used, the setting information for the machine or equipment, the operation details of the machine or equipment, the anomaly detection mode, the type of data to be used for anomaly detection, the model to be used for anomaly detection, and so on.

[0021] On the anomaly detection condition selection screen in Figure 3, you can enter conditions such as "machining center" as the type of machine tool, "end mill" as the part to be used, and "tapping" as the operation.

[0022] Anomaly detection modes include "Emphasis on accuracy" and "Emphasis on speed." Selecting an anomaly detection mode determines the anomaly detection model and type of data to be used for anomaly detection that are appropriate for the selected mode. The types of data used for anomaly detection include current voltage, temperature, humidity, infrared, acceleration, magnetism, pressure, tilt, fluid velocity, vibration, rotational speed, and torque. For example, when detecting an abnormality in a motor or a moving part caused by the motor, data such as torque command, vibration, and sound is used. When "speed-oriented" is selected as the anomaly detection mode, a model that detects anomalies using one type or a small number of data (for example, torque commands) is selected. To increase accuracy, a model that combines multiple or many types of data in addition to torque commands is selected.

[0023] Figure 4 shows the difference in the number of data points between "accuracy-oriented" and "speed-oriented". In "accuracy-oriented" signal processing, which processes multiple pieces of data, multiple signals (signal A to signal N) are input for each time block. In "speed-oriented" signal processing, one signal (signal A) is input for each time block. The anomaly detection unit 13 processes the input data sequentially, but the smaller the number of pieces of data, the faster the processing speed.

[0024] Anomaly detection models include the MT method and variational auto-encoder (VAE). The MT method is a method for quantifying whether a target falls within a standard when normal data is used as the standard. Methods like the MT method that calculate the degree of abnormality using the average value or standard deviation have a low computational load. Variational autoencoders are a type of deep learning model that can improve accuracy when large amounts of well-organized data are present, but are computationally intensive. When selecting the anomaly detection mode, if you select "speed priority," a model with low computational load, such as the MT method, will be selected. If you select "accuracy priority," a model with high computational load but relatively high accuracy, such as VAE, will be selected.

[0025] The operator can also directly select the model to be used for anomaly detection. The selection screen in Figure 3 provides selection areas for the model to be used for anomaly detection and the type of data to be used for anomaly detection.

[0026] If the conditions for anomaly detection change, the waveform characteristics and trends will be different. The anomaly detection unit 13 and the anomaly analysis unit 17 store the results of anomaly detection and anomaly analysis in the diagnosis history storage unit 20. The diagnosis history storage unit 20 also stores models that have been created in advance by learning for each condition.

[0027] The abnormality notification unit 15 monitors the abnormality detection data and notifies the operator when an abnormality is detected. Figure 5 is an example of an abnormality notification screen. This screen displays the abnormality level of the data used for abnormality detection (here, torque command) "torque command abnormality level: 10.1" and the abnormality probability for the abnormality level "machine abnormality probability 96.1%". The graph in Figure 5 shows the progress of the machine abnormality probability. The abnormality notification screen is always displayed. The operator can check the results of abnormality detection in real time. When an abnormality occurs, the abnormality notification unit 15 displays the abnormality level of the abnormality analysis data on the abnormality notification screen. The operator can analyze the cause of the abnormality based on the abnormality levels of multiple data.

[0028] The data extraction unit 16 extracts data based on the results of the anomaly detection. In Figure 6, the upper part shows the motor torque command under normal conditions, and the lower part shows the motor torque command under abnormal conditions. The area between the dotted lines in the figure indicates the extraction range. The extraction range is determined according to a predetermined rule. As an example, a predetermined amount of data is extracted from before and after the point in time when the anomaly is detected. There are no particular restrictions on the extraction rule. Figure 6 shows that the waveforms of the extracted data are significantly different between normal and abnormal conditions.

[0029] The data extraction unit 16 also extracts data to be used for abnormality analysis based on the results of abnormality detection. Figure 7 shows the extraction ranges for Z-axis position feedback, cutting signal, and rotation speed. The data extraction unit also extracts other data based on the results of abnormality detection.

[0030] The fault analysis unit 17 acquires the data extracted by the data extraction unit 16 and analyzes the cause of the fault. The diagnosis history storage unit 20 stores a fault analysis model generated from the previously extracted data. The fault analysis unit 17 uses the fault analysis model to analyze the cause of the fault from the extracted data.

[0031] The analysis result notification unit 18 notifies the operator of the results of the abnormality analysis. Fig. 8 shows an example of a display screen for the analysis results of the abnormality cause. The display screen in Fig. 8 shows the analysis results in probabilities, such as "Probability of tool breakage: 91.1%", "Probability of chip clogging: 6%", "Probability of spindle grease shortage: 2%", "Probability of spindle damage: 0.1%", and "Probability of sequence number discrepancy: 0.001%".

[0032] The anomaly analysis condition selection unit 19 accepts the selection of anomaly analysis conditions by the operator. By setting the anomaly analysis conditions, an appropriate model can be selected. Figure 9 shows an example of the anomaly analysis condition selection screen. On the selection screen, the anomaly analysis mode, the type of data to be used in anomaly analysis, the model to be used in anomaly analysis, etc. can be selected.

[0033] The diagnostic history storage unit 20 stores the conditions for anomaly detection, the results of anomaly detection, the extracted data, the conditions for anomaly analysis, the results of anomaly analysis, a model for anomaly detection, a model for anomaly analysis, etc. Past diagnostic history is used for model construction, anomaly detection, and anomaly analysis. The diagnostic history presentation unit 21 presents to the operator the past diagnostic history stored in the diagnostic history storage unit 20. The past diagnostic history can be used when selecting conditions for anomaly detection and anomaly analysis. The past diagnostic history serves as a reference when selecting the type of model and data to be used in anomaly detection and anomaly analysis.

[0034] The operation of the abnormality diagnostic device 100 of the first disclosure will be described with reference to the flowchart of FIG. The fault diagnosis device 100 acquires fault detection conditions from an operator (step S1). The fault diagnosis device 100 displays, for example, a selection screen for fault detection conditions and accepts input from the operator. On the selection screen, the fault detection mode, the type of data to be used for fault detection, the model to be used for fault detection, etc. can be selected.

[0035] The fault diagnosis device 100 acquires fault analysis conditions from the operator (step S2). The fault diagnosis device 100 displays, for example, a fault analysis conditions selection screen and accepts input from the operator. On the selection screen, the fault analysis mode, the type of data to be used in the fault analysis, the model to be used in the fault analysis, etc. can be selected. Note that the fault analysis conditions may be selected before the fault analysis in step S7.

[0036] The fault diagnosis device 100 acquires data from the equipment and machines in the factory and from sensors installed in the factory. The fault diagnosis device 100 performs fault detection under the conditions selected in step S1 (step S3). The fault detection in step S3 is performed in real time using less data than in the fault analysis.

[0037] When the occurrence of an abnormality is detected (step S4; Yes), the abnormality diagnosis device 100 notifies the operator of the occurrence of the abnormality (step S5). The abnormality diagnosis device 100 displays, for example, an abnormality notification screen, which displays the degree of abnormality of the data that is the target of abnormality detection and the probability that the abnormality has occurred.

[0038] If the occurrence of an abnormality is not detected (step S4; No), the process proceeds to step S3, and abnormality detection continues.

[0039] When an abnormality is detected, the abnormality diagnostic device 100 extracts data to be used for analyzing the abnormality (step S6). The type of data to be used for the abnormality analysis is selected in step S2. The range of data extraction is determined according to a predetermined rule.

[0040] The abnormality diagnosis device 100 uses the extracted data to perform an abnormality analysis under the conditions selected in step S2 (step S7). The abnormality analysis in step S7 does not need to be performed in real time. A larger amount of data than the abnormality detection in step S3 is used to infer the location of the abnormality, the cause of the abnormality, etc.

[0041] The fault diagnosis device 100 notifies the operator of the results of the fault analysis (step S8). The fault diagnosis device 100 displays the results of the fault analysis, for example, on a display screen. The display screen displays the estimated causes of the fault and their probabilities. The operator uses the results of the fault analysis as a reference to identify the actual cause of the fault. The operator's confirmation results are fed back to the fault diagnosis device as the true cause of the fault.

[0042] The abnormality diagnosis device 100 records the conditions for abnormality detection, the model for abnormality detection, the results of abnormality detection, the extracted data, the conditions for abnormality analysis, the results of abnormality analysis, the model for abnormality analysis, the true cause of the abnormality as fed back by the operator, etc. in the diagnosis history storage unit (step S9).

[0043] As described above, the anomaly diagnosis device 100 of the first disclosure divides anomaly diagnosis into two stages: anomaly detection and anomaly analysis. The model used for anomaly detection uses less data and has a lower calculation load than the model used for anomaly analysis. This reduces the data acquisition load and calculation load in anomaly detection, allowing anomaly detection to be performed without delay. The first and second disclosed fault diagnosis devices 100 hold multiple models capable of identifying the cause of a fault. When a fault occurs, fault diagnosis data before and after the fault, including the time point at which the fault occurred, are extracted, and the degree of deviation from normal data is calculated to identify the cause of the fault. The conditions for anomaly analysis can be set by the operator. The operator can set an appropriate diagnostic method while taking into consideration factors such as the computational load, communication load, and whether to prioritize accuracy or speed. According to the present disclosure, it is possible to improve factory productivity and reliability while suppressing computational costs, communication costs, etc.

[0044] The abnormality diagnostic device 100 displays the abnormality level, which is the result of the abnormality detection, in real time. The operator can monitor the change in the abnormality level and predict the occurrence of an abnormality.

[0045] The anomaly diagnosis device 100 records the anomaly detection conditions, the anomaly detection results, the estimated accuracy of the anomaly detection, the anomaly analysis conditions, the anomaly analysis results, the estimated accuracy of the anomaly analysis, and the extracted data. The operator can select an appropriate model based on the estimated accuracy of past anomaly diagnoses, etc.

[0046] [Second Disclosure] The second disclosure will be described. The abnormality diagnosis system 200 of the second disclosure includes the first Disclosure The components of the anomaly diagnosis device are distributed across a factory system. Figure 11 shows an example of a factory system. As shown in Figure 11, the factory system is composed of edge, fog, and cloud. The edge is an area close to sensors and control devices in terms of the network. Edge computing performs real-time processing of large volumes of data obtained from numerous sensors installed in factories and from the control devices of factory equipment and machinery. The cloud is not an internal hard drive or local server, but an external system accessed via the internet. Cloud computing is the process of integrating large amounts of data collected at the edge. of The data is stored in the cloud and analyzed. The fog is located between the cloud and the edge in terms of the network. The fog may perform data processing that would normally be performed in the cloud. Because the data is processed before being sent over the Internet, it can respond quickly to changes in the environment.

[0047] In the abnormality diagnosis system 200 of the second disclosure, the abnormality detection data acquisition unit 11 and the abnormality detection unit 13 are implemented on an edge or a fog. Specifically, the abnormality detection data acquisition unit 11 and the abnormality notification unit 15 are implemented in an information processing device (for example, a computerized numerical control (CNC), a PLC, a local server) that is close to the edge in terms of the network, or in an information processing device (for example, a gateway) that is close to the fog in terms of the network. The abnormality analysis data acquisition unit 12, the data extraction unit 16, and the abnormality analysis unit 17 are implemented in an information processing device on the cloud or an information processing device close to the fog. The diagnosis history storage unit 20 is implemented in a storage device on the cloud. The anomaly detection condition selection unit 14, the anomaly analysis condition selection unit 19, and the analysis result notification unit 18 may be implemented in any of the edge, fog, and cloud.

[0048] In the abnormality diagnosis system 200 of the second disclosure, abnormality detection is performed by an information processing device close to the edge or fog, and abnormality analysis is performed by an information processing device close to the fog or the cloud. In the abnormality diagnosis system, abnormality detection conditions and abnormality analysis conditions are selected in advance, and an appropriate abnormality diagnosis method can be set while taking into consideration factors such as the calculation load, communication load, and whether to prioritize accuracy or speed. The anomaly diagnosis system can detect anomalies in real time at the edge or in the fog, and analyze the data in which anomalies are detected in the fog or in the cloud. [Explanation of symbols]

[0049] 100 Abnormality diagnosis device 200 Abnormality diagnosis system 11 Anomaly detection data acquisition unit 12. Abnormality analysis data acquisition section 13 Abnormality detection unit 14. Anomaly detection condition selection section 15 Abnormality notification section 16 Data extraction section 17 Anomaly Analysis Department 18 Analysis result notification department 19. Anomaly analysis condition selection section 20 Diagnostic history memory section 21 Diagnostic history presentation section 111 CPU 112 ROM 113 RAM 114 Non-volatile memory

Claims

1. An abnormality diagnosis device for diagnosing abnormalities occurring in a factory, an anomaly detection data acquisition unit that acquires data used for anomaly detection; an anomaly detection condition selection unit that accepts selection of data to be used in the anomaly detection and an anomaly detection condition to be used in the anomaly detection; an anomaly detection unit that detects an anomaly of a diagnosis target using the data acquired by the anomaly detection data acquisition unit; an abnormality notification unit that notifies the abnormality detected by the abnormality detection unit; an abnormality analysis data acquisition unit that acquires data used for abnormality analysis; an anomaly analysis unit that analyzes candidates for anomaly factors using data used for the anomaly analysis during a period including the time when the anomaly occurred, or data used for the anomaly detection during a period including the time when the anomaly was detected and data used for the anomaly analysis; Equipped with the anomaly detection condition selection unit accepts a selection of whether emphasis is placed on accuracy or speed in the anomaly detection, and determines data to be used for the anomaly detection and a model to be used for the anomaly detection based on the selection. Abnormality diagnosis device.

2. The anomaly detection unit identifies a feature amount of data related to an anomaly based on data at normal times and data at abnormal times, The abnormality diagnosis device according to claim 1 , wherein the abnormality notification unit notifies the characteristic amount.

3. the anomaly detection unit detects an anomaly degree for each of the feature amounts, The abnormality diagnosis device according to claim 2 , wherein the abnormality notification unit notifies the feature quantity having a high degree of abnormality.

4. 2. The abnormality diagnosis device according to claim 1, further comprising an abnormality analysis condition selection unit that accepts selection of data to be used in the abnormality analysis and abnormality analysis conditions to be used in the abnormality analysis.

5. The abnormality diagnosis device according to claim 1 , wherein the abnormality notification unit notifies the degree of abnormality of the data used for the abnormality detection in real time.

6. a diagnosis history storage unit that stores the results of the abnormality diagnosis; a diagnosis history presentation unit that presents the results of the abnormality diagnosis; The abnormality diagnosis device according to claim 1 , comprising:

7. An abnormality diagnosis system for diagnosing abnormalities occurring in a factory, an anomaly detection data acquisition unit that acquires data used for anomaly detection; an anomaly detection condition selection unit that accepts selection of data to be used in the anomaly detection and an anomaly detection condition to be used in the anomaly detection; an anomaly detection unit that detects an anomaly of a diagnosis target using the data acquired by the anomaly detection data acquisition unit; an abnormality notification unit that notifies the abnormality detected by the abnormality detection unit; an abnormality analysis data acquisition unit that acquires data used for abnormality analysis; an anomaly analysis unit that analyzes candidates for anomaly factors using data used for the anomaly analysis during a period including the time when the anomaly occurred, or data used for the anomaly detection during a period including the time when the anomaly was detected and data used for the anomaly analysis; Equipped with the anomaly detection condition selection unit accepts a selection of whether emphasis is placed on accuracy or speed in the anomaly detection, and determines data to be used for the anomaly detection and a model to be used for the anomaly detection based on the selection. Abnormality diagnosis system.

8. By executing one or more processors, the device operates as an anomaly detection data acquisition unit, an anomaly detection condition selection unit, an anomaly detection unit, an anomaly notification unit, an anomaly analysis data acquisition unit, and an anomaly analysis unit, the anomaly detection data acquisition unit acquires data used for detecting anomalies occurring in the factory, the anomaly detection condition selection unit accepts a selection of whether to prioritize accuracy or speed in the anomaly detection of the anomaly, and determines data to be used for the anomaly detection and a model to be used for the anomaly detection based on the selection; the anomaly detection unit detects an anomaly of a diagnosis target using the data used for the anomaly detection, the abnormality notification unit notifies the detected abnormality, the abnormality analysis data acquisition unit acquires data to be used for abnormality analysis, the anomaly analysis unit analyzes candidate anomaly factors using data used for the anomaly analysis for a period including the time point at which the anomaly occurred, or data used for the anomaly detection for a period including the time point at which the anomaly was detected and data used for the anomaly analysis; A computer-readable storage medium that stores instructions.

Images