Abnormality diagnosis system and abnormality diagnosis method

Abstract

In this invention, the data analysis unit has a suitability judgment unit, which calculates a judgment factor for judging the suitability of the data analysis process to judge the suitability of the data analysis process, and the display device displays the suitability judgment result.

Description

Technical Field

[0001] This invention relates to an anomaly diagnosis system and anomaly diagnosis method. Background Technology

[0002] In equipment management on production lines in manufacturing plants, proper maintenance is crucial for continuous production. Therefore, to avoid the need to halt equipment operation due to unexpected malfunctions, technologies for detecting equipment anomalies have been developed to prevent failures.

[0003] In recent years, anomaly detection technologies have been developed that utilize information from equipment status monitoring and analyze data such as equipment learning to detect anomalies. To detect faults accurately and proactively, the appropriateness of the data analysis process needs to be thoroughly validated.

[0004] As a method for verifying the appropriateness of the data analysis process in such anomaly detection, as described in Patent Document 1, there exists a system that uses graphical displays such as histograms of state monitoring data to assist analysts in judging its appropriateness.

[0005] Existing technical documents

[0006] Patent documents

[0007] Patent Document 1: Japanese Patent Application Publication No. 2019-16209 Summary of the Invention

[0008] The problem that the invention aims to solve

[0009] When verifying the appropriateness of the data analysis process for anomaly diagnosis based on histograms and other data from status monitoring, it is necessary to determine what kind of display is appropriate. Therefore, expertise in data analysis and expertise in the use of the equipment are indispensable.

[0010] However, while engineers at the field possess expertise in the equipment itself and the symptoms of failure, very few have expertise in data analysis.

[0011] The purpose of this invention is to provide an anomaly diagnosis system and method that allows even those without expertise in data analysis to easily determine the appropriateness of a data analysis process.

[0012] Technical solutions for solving the problem

[0013] An anomaly diagnosis system according to one aspect of the present invention is an anomaly diagnosis system for diagnosing anomalies in industrial equipment, characterized in that it includes: a data analysis unit for analyzing data based on status monitoring information of the industrial equipment; and a display device for displaying the analysis results of the data, wherein the data analysis unit has an appropriateness judgment unit that calculates a judgment factor for judging the appropriateness of the data analysis process to judge the appropriateness of the data analysis process, and the display device displays the appropriateness judgment result.

[0014] One aspect of the present invention is an anomaly diagnosis method for diagnosing anomalies in industrial equipment, characterized in that: data is analyzed based on the status monitoring information of the industrial equipment, a judgment factor is calculated to determine the appropriateness of the data analysis process, and the analysis results and the appropriateness judgment results are displayed.

[0015] Invention Effects

[0016] According to one aspect of the present invention, even those without expertise in data analysis can easily determine the appropriateness of a data analysis process. Attached Figure Description

[0017] Figure 1 This is a diagram showing the structure of the anomaly diagnosis system of Embodiment 1.

[0018] Figure 2 This is a diagram illustrating an example of the user interface of Embodiment 1.

[0019] Figure 3 This is a diagram representing an example of the decision factor in Example 1.

[0020] Figure 4 This is a diagram showing the structure of the anomaly diagnosis system of Embodiment 2.

[0021] Figure 5 This is a diagram illustrating an example of the user interface of Embodiment 2.

[0022] Figure 6 This is a diagram showing the structure of the anomaly diagnosis system in Example 3.

[0023] Figure 7 This is a diagram illustrating an example of the user interface of Embodiment 3.

[0024] Figure 8 This is a diagram showing the structure of the anomaly diagnosis system in Example 4.

[0025] Figure 9 This is a diagram showing the structure of the anomaly diagnosis system of Example 5. Detailed Implementation

[0026] The embodiments are described below using the accompanying drawings.

[0027] Example 1

[0028] refer to Figure 1 The structure of the anomaly diagnosis system in Example 1 will be described.

[0029] The status monitoring unit 101 monitors the status of industrial equipment, and the data analysis unit 102 uses this information to detect abnormalities and deterioration trends of the equipment, and displays the analysis results on the display input device 107.

[0030] The data analysis unit 102 includes a preprocessing unit 103 that processes data sent from the status monitoring unit 101 into data required for analysis, a feature extraction unit 104 that extracts information as features of the analysis from the preprocessed data, a diagnostic unit 106 that uses the extracted feature quantities to deduce the abnormality and deterioration tendency of the equipment, and an appropriateness judgment unit 113 that judges the appropriateness of the data analysis.

[0031] The suitability judgment unit 113 calculates a judgment factor for judging the suitability of data analysis based on the input and output information of the preprocessing unit 103 and the feature extraction unit 104, judges the suitability of data analysis based on the calculated judgment factor, and displays the suitability judgment result 109 on the display input device 107.

[0032] To perform the diagnosis, diagnostic parameters 108 are input from the display input device 107. In the case of multiple diagnoses, the diagnostic model 105 is specified as the diagnostic parameters 108. In addition, the determination database 110 stores correction methods for the diagnostic parameters 108 corresponding to the suitability judgment result 109, and sends the appropriate correction method for the diagnostic parameters 108 to the display input device 107 in accordance with the suitability judgment result 109.

[0033] Figure 2 This represents an example of a user interface displayed on the display input device 107.

[0034] The upper section shows the interface for the preprocessing unit 103, feature extraction unit 104, diagnostic model 105, and their corresponding diagnostic parameters 108 used in input data analysis. In this example, FFT (Fast Fourier Transform) is selected as the preprocessing method, higher harmonics as the feature quantity, and MT method (Mathau's Taguchi method) as the diagnostic method. Furthermore, initial values ​​are given for two parameters A and B for FFT, and initial values ​​are given for three parameters X, Y, and Z for higher harmonics.

[0035] Based on this input information and data from the condition monitoring unit 101, the diagnostic unit 106 diagnoses the abnormalities and deterioration tendencies of the industrial equipment and displays the analysis results in the user interface at the bottom.

[0036] In addition, the lower user interface displays, in addition to the analysis results, the appropriateness judgment result 109 and the correction method for the diagnostic parameters corresponding to the appropriateness judgment result 109 stored in the judgment database 110.

[0037] In this example, three judgment items (1), (2), and (3) are shown, where one judgment item (1) is OK (appropriate analysis has been performed) and the other two judgment items (2) and (3) are NG (appropriate analysis has not been performed).

[0038] Furthermore, correction methods are shown for each of the NG (Not Given) judgment items. Specifically, for judgment item (2), an indication is shown to increase the measurement range of the state monitoring. Additionally, for judgment item (3), an indication is shown to increase the parameter X of the higher harmonics.

[0039] Thus, the suitability judgment result 109 is displayed on the display input device 107 to indicate whether the data analysis is appropriate. Furthermore, the correction method for the diagnostic parameter 108 is displayed on the display input device 107 as a correction method. Therefore, even a person without expertise in data analysis can appropriately adjust the diagnostic parameter 108.

[0040] Next, in Figure 3 The example shown is a derived example of the appropriateness judgment result 109.

[0041] For the two judgment item examples, examples of using graphics to judge the appropriateness of the data analysis process and examples of calculating the judgment factors are shown respectively. First, in Figure 3 Example of appropriateness judgment regarding the measurement range of the status monitoring unit 101 is shown in (1) in relation to the preprocessing of data analysis.

[0042] Condition monitoring should be conducted within the measurement range. If the condition exceeds this range, the correct condition quantity cannot be measured, thus hindering accurate anomaly diagnosis. Therefore, using graphical methods to assess suitability allows for judgment based on whether the condition monitoring signal is within the measurement range.

[0043] When calculating this judgment as a decision factor, the decision factor is considered appropriately calculated if the maximum value of the status monitoring signal is less than the upper limit of the measurement interval and the minimum value of the signal is greater than the lower limit of the measurement interval. On the other hand, the decision factor is considered not appropriately calculated if the maximum value of the signal is the same as the upper limit of the measurement interval or the minimum value of the signal is the same as the lower limit of the measurement interval.

[0044] In this way, the appropriateness of data analysis can be judged using automatically calculated decision factors without the need for graphical confirmation of status monitoring signals.

[0045] Next, in Figure 3 Example of the appropriateness judgment for peak detection in feature extraction is shown in (2).

[0046] When extracting features from frequency spectra, there are cases where a peak value is extracted as a feature value from a certain detection amplitude. In this case, to determine the appropriateness of the feature value graphically, it is considered appropriate if the detected peak value is significantly larger than the noise level of the surrounding spectrum, and not appropriate if it is equal to the noise level.

[0047] When this judgment is calculated as a decision factor, appropriateness is determined if the value of (peak value / average value of surrounding signals) is greater than a decision threshold, and inappropriateness is determined if it is less than the threshold. For example, if the decision threshold is 2, appropriateness is determined if the peak value is more than twice the average value of surrounding signals, and inappropriateness is determined if it is less than twice the average value of surrounding signals. This decision threshold can be predetermined or given as part of diagnostic parameter 108.

[0048] In this way, the appropriateness of peak detection can be determined using decision factors instead of graphical methods. Multiple decision factors can be calculated based on the data analysis process, or multiple decision factors can be calculated from a single data analysis process.

[0049] Next, with Figure 3 Similarly, regarding peak detection in feature extraction, the appropriateness of representing the position of the detected peak also needs to be determined. Figure 3 Example of the appropriateness judgment regarding the peak position is shown in (2).

[0050] As an example of graphical judgment, for a given peak detection range, if the actual detected peak position is within the detection range, the data analysis can be judged as appropriate. If the detected peak position is at the end of the detection range, there is a possibility that a larger peak exists outside the detection range, which can be judged as inappropriate data analysis. Using this judgment as a factor, the peak position is compared with the upper and lower limits of the detection range. If the peak position is between the upper and lower limits, it can be judged as appropriate; if the peak position is the same as the upper and lower limits, it can be judged as inappropriate.

[0051] In this way, multiple suitability judgment results may be calculated based on one data analysis process 109.

[0052] In addition, for Figure 3For example 109 of the appropriateness judgment results of (2) and (3), we consider expanding the measurement interval and expanding the peak detection amplitude as correction methods for data analysis in cases where the appropriateness is not judged.

[0053] Therefore, by storing such a combination of the correctable diagnostic parameter 108 and the suitability judgment result 109 in the judgment database 110, a correction method for the diagnostic parameter 108 can be derived based on the suitability judgment result 109, and the result can be displayed on the display input device 107.

[0054] However, the display of such suitability judgment results 109 and the input of diagnostic parameters 108 can also be achieved using not only methods such as... Figure 2 The graphical user interface shown also uses the program's source code and graphical display functions.

[0055] Example 2

[0056] refer to Figure 4 The structure of the anomaly diagnosis system in Example 2 will be described.

[0057] Typically, when adjusting the diagnostic model 105, as shown in Example 1, the diagnostic parameter 108 is changed for adjustment. In Example 2, the adjustment range of the diagnostic parameter 108 is given in advance as the diagnostic parameter range 111, in which the diagnosis is performed and the suitability judgment result 109 is displayed on the display input device 107. Thus, the diagnostic parameter 108 can be adjusted quickly.

[0058] However, it is not necessary to input the diagnostic parameter search range 111 for all diagnostic parameters 108; diagnostic parameters 108 can be used simultaneously. Additionally, in Example 2, there is no... Figure 1 The determination database 110 is shown. Other structures are similar to... Figure 1 The abnormality diagnosis system shown in Embodiment 1 has the same structure, so its description is omitted.

[0059] exist Figure 5 The image shows an example of the user interface when both diagnostic parameter 108 and diagnostic parameter search amplitude 111 are used simultaneously. The selection is related to... Figure 2 The same preprocessing, feature quantity, and diagnosis are used. In the parameters, the diagnostic parameter search range 111 is used for parameter X, and the diagnostic parameter 108 is specified for all other parameters.

[0060] At this point, during the diagnostic process, multiple data analyses are performed corresponding to the diagnostic parameter search range 111, and the decision factors from each data analysis are displayed together in the middle. In this way, by displaying the decision factors corresponding to the diagnostic parameter search range 111 together, it is possible to select as many parameters as possible that have undergone data analysis as appropriate, and to quickly adjust the diagnostic parameter 108.

[0061] Alternatively, the diagnostic parameter search range 111 can be adjusted sequentially for each parameter, or the diagnostic parameter search range 111 can be set and adjusted simultaneously for multiple parameters.

[0062] Example 3

[0063] refer to Figure 6 The structure of the anomaly diagnosis system in Example 3 will be described.

[0064] In Example 3, for the data analysis unit 102, a diagnostic package 112, which combines preprocessing, feature extraction, and diagnosis, is used to reduce the number of diagnostic parameters 108 input from the display input device 107, thereby making it easier to adjust the diagnostic parameters. Other structures are similar to... Figure 1 The abnormality diagnosis system shown in Embodiment 1 has the same structure, so its description is omitted.

[0065] At this point, diagnostic package 112 may not be a single package, but rather multiple diagnostic packages 112 may be prepared to detect several faults. Figure 7 An example of the user interface in this case is shown below.

[0066] Figure 7 Three diagnostic packages were prepared, and two of them, degradation A and degradation B, were selected, with their corresponding parameters entered. When the diagnosis was executed, the diagnostic results corresponding to each diagnostic package 112 were displayed, and the appropriateness judgment results 109 were also displayed for each diagnostic package 112.

[0067] In this way, diagnostic parameters 108 can be easily adjusted without having to choose a combination of preprocessing, feature extraction, and diagnosis.

[0068] Example 4

[0069] refer to Figure 8 The structure of the anomaly diagnosis system in Example 4 will be described.

[0070] In Example 4, as Figure 8 As shown, for the diagnostic target device 100 owned by the equipment user, the diagnostic parameters 108 are adjusted by the system provider of the anomaly detection system. The status of the diagnostic target device 100 is monitored by the status monitoring unit 101, and based on the monitored data, the data analysis unit 102 detects the tendency of anomalies and deterioration, and sends the results to the display input device 107.

[0071] At this time, all parts except the diagnostic device 100 and the status monitoring unit 101 can be shared units that can be accessed by both the device user and the system provider. The suitability judgment result 109 is sent to the system provider from the data analysis unit 102, and the system provider adjusts the diagnostic parameters 108 based on its own judgment database 110.

[0072] Therefore, the system provider can also remotely adjust the diagnostic parameters 108 appropriately. Additionally, the equipment user can also adjust the diagnostic parameters 108 using the common components, only entrusting the system provider to adjust them when adjustment is difficult. As a result, by charging for the amount of adjustment entrusted to the system provider, a usage fee corresponding to the system provider's workload can be calculated.

[0073] Example 5

[0074] refer to Figure 9 The structure of the anomaly diagnosis system in Example 5 will be described.

[0075] In Example 5, as Figure 9 As shown, a diagnostic device 120 is used, which integrates the status monitoring unit 101 and the data analysis unit 102. The diagnostic device 120, which is connected to the diagnostic target device 100, has a status monitoring unit 101 and a data analysis unit 102. It performs data analysis in response to diagnostic parameters 108 sent from the network GUI 121 and sends the data analysis results to the network GUI 121 for display and input.

[0076] The network GUI 121 can be built into the diagnostic device 120 or configured in a network. In this case, by sending the suitability judgment result 109 from the diagnostic device 120 to the memory 122 and comparing it with the judgment database 110 stored in the memory 122, the correction method of the diagnostic parameter 108 is displayed on the network GUI 121.

[0077] In this way, the analyst can use the network GUI121 to appropriately and easily modify the diagnostic parameter 108.

[0078] According to the above embodiments, even people without expertise in data analysis can easily determine the appropriateness of the data analysis process.

[0079] Here, Figure 1 The data analysis unit 102 shown is, for example, a computer equipped with a storage device and a CPU. Furthermore, Figure 1 The “~ part” shown, for example, achieves its “function” by executing a program by a processor (CPU, etc.).

[0080] Explanation of reference numerals in the attached figures

[0081] 100 Diagnostic Devices

[0082] 101 Status Monitoring Department

[0083] 102 Data Analysis Department

[0084] 103 Pre-processing Department

[0085] 104 Feature Extraction Section

[0086] 105 Diagnostic Model

[0087] 106 Diagnostic Department

[0088] 107 Display Input Device

[0089] 108 Diagnostic Parameters

[0090] 109 Appropriateness Judgment Results

[0091] 110 Judgment Database

[0092] 111 Diagnostic Parameter Search Range

[0093] 112 Diagnostic Kit

[0094] 113 Appropriateness Judgment Department

[0095] 120 Diagnostic Device

[0096] 121 Display / Input Network GUI

[0097] 122 Memory.

Claims

1. An abnormality diagnosis system that diagnoses an abnormality of an industrial device, characterized by, comprises: a data analysis section that performs analysis of data based on state monitoring information of the industrial equipment; and a display device constituted by a display input device that can input prescribed information, which displays a result of the analysis of the data, the data analysis section has: a propriety judgment section that calculates a judgment factor for judging propriety of an analysis process of the data to judge the propriety of the analysis process of the data, a preprocessing section that pre-processes the data required for the analysis of the state monitoring information for processing, a feature quantity extraction section that extracts a feature quantity that is a feature of the analysis from the data on which the preprocessing has been performed, and a diagnosis section that diagnoses an abnormality of the industrial equipment using the feature quantity, the propriety judgment section calculates the judgment factor for judging the propriety based on input information from the preprocessing section and the feature quantity extraction section, the diagnosis section diagnoses an abnormality of the industrial equipment using a diagnosis parameter and a diagnosis model input from the display input device, the abnormality diagnosis system further has a judgment database that stores a correction method of the diagnosis parameter corresponding to a judgment result of the propriety, a correction method of the diagnosis parameter corresponding to the judgment result of the propriety is read from the judgment database and displayed on the display device, the display device, displays, as the judgment result of the propriety, whether the analysis of the data was performed properly for each judgment item, in a case where the analysis of the data was not performed properly, a correction method of the diagnosis parameter corresponding to the judgment result of the propriety is read from the judgment database, the correction method of the diagnosis parameter is displayed corresponding to the judgment item for which the analysis of the data was not performed properly.

2. The abnormality diagnosis system according to claim 1, wherein: the propriety judgment section sets the judgment factor in a manner that the state monitoring information converges within a prescribed measurement interval range.

3. The abnormality diagnosis system according to claim 1, wherein: the propriety judgment section sets the judgment factor in a manner that a peak value or a peak position in the feature quantity converges within a prescribed range.

4. The abnormality diagnosis system according to claim 1, wherein: the diagnosis section, in a case where the diagnosis parameter is to be changed to adjust the diagnosis model, diagnoses an abnormality of the industrial equipment from the display input device by previously inputting an adjustment range of the diagnosis parameter as a diagnosis parameter search range, the data analysis section performs the analysis of the data a plurality of times in accordance with the diagnosis parameter search range, the display input device displays the judgment result of the propriety corresponding to the diagnosis parameter search range.

5. The abnormality diagnosis system according to claim 1, wherein: the data analysis section is constituted by a plurality of diagnosis packs, and the propriety of the analysis process of the data is judged for each of the diagnosis packs, the display device displays the judgment result of the propriety for each of the diagnosis packs.

6. The abnormality diagnosis system according to claim 1, wherein: sending the decision factor from an owner of the industrial equipment to a system provider of the abnormality diagnosis system, a fee of the abnormality diagnosis system changes depending on a number or amount of times the system provider adjusts a diagnosis parameter used for the analysis of the data.

7. The abnormality diagnosis system according to claim 1, wherein: a diagnosis device having a state monitoring unit that monitors a state of the industrial equipment and the data analysis unit, a storage that has a judgment database in which a correction method of a diagnosis parameter used for the analysis of the data is stored in correspondence with a result of the judgment of the appropriateness, and the display device that displays a result of the analysis of the data are connected by a network, the display device displays the result of the judgment of the appropriateness of the analysis process of the data and the correction method of the diagnosis parameter corresponding to the result of the judgment of the appropriateness read from the judgment database.

8. An abnormality diagnosis method of diagnosing an abnormality of an industrial equipment, characterized by: performing an analysis of data based on state monitoring information of the industrial equipment, calculating a decision factor for judging appropriateness of the analysis process of the data to judge the appropriateness of the analysis process of the data, preprocessing the data required for the analysis of the state monitoring information for processing, extracting a feature amount as a feature of the analysis from the data on which the preprocessing is performed, diagnosing an abnormality of the industrial equipment using the feature amount, calculating the decision factor for judging the appropriateness based on information obtained by performing the preprocessing and the extraction of the feature amount, diagnosing an abnormality of the industrial equipment using an input diagnosis parameter and a diagnosis model, reading and displaying a correction method of the diagnosis parameter corresponding to a result of the judgment of the appropriateness from a judgment database in which the correction method of the diagnosis parameter corresponding to the result of the judgment of the appropriateness is stored, displaying a result of the analysis of the data and displaying whether the analysis of the data is appropriately performed as the result of the judgment of the appropriateness for each judgment item, in a case where the analysis of the data is not appropriately performed, reading a correction method of the diagnosis parameter corresponding to a result of the judgment of the appropriateness from the judgment database, displaying the correction method of the diagnosis parameter corresponding to the judgment item in which the analysis of the data is not appropriately performed.

Images