Method and system for evaluating a technical plant

Abstract

A method for evaluating a technical plant is proposed. An operating data set is recorded during an operating phase. The operating data set is assigned to a class by a classifier. At least one performance indicator is generated from the operating data set. The generated performance indicator is assigned to the operating data set. The performance indictor assigned to the operating data set is assessed. A system according to the invention for evaluating a technical plant comprises a data memory and a classifier. The system according to the invention is configured to carry out the method according to the invention.

Description

[0001] Procedure and system for evaluating a technical plant

[0002] Description:

[0003] The invention relates to a method for evaluating a technical plant, wherein an operational data set is recorded during an operating phase and the operational data set is assigned to a class by a classifier. The invention also relates to a system for evaluating a technical plant, which is configured to carry out the method according to the invention.

[0004] Technical systems, such as drive systems, comprise a multitude of technical components, including an electric motor, a converter to generate three-phase alternating current for the electric motor, and a gearbox to reduce the motor's speed. Examples of such technical systems include rotary tables, conveyor belts, and stacker cranes. After extended operation, the components of these systems can malfunction due to wear and tear.

[0005] It is known to monitor such technical systems by recording and evaluating specific measurements at set times. If the recorded measurements deviate too significantly from predefined target values, a fault in the system is assumed, and a corresponding message is sent to the operator. For example, a current generated by the inverter that exceeds a defined limit could indicate stiffness in the gearbox due to wear.

[0006] In normal operation, the technical system moves, for example, a light load and a heavy load. To detect stiffness, the current is monitored against a limit value. This limit value must be defined such that it is not exceeded when the heavy load is applied. If stiffness occurs in the technical system under a light load, the limit value will not be exceeded, and the stiffness cannot be detected. A method for detecting faults and assigning them to their causes in a hydrostatic system, as well as a corresponding control device, are disclosed in DE 102008 063 924 A1.

[0007] From DE 102013 100411 A1 a method and a device for condition monitoring of a technical plant are known.

[0008] From DE 102022 114 579 A1 an industrial plant and a method for plant operation and plant monitoring are known.

[0009] A method for condition analysis of a technical plant is known from WO 2020 / 216452 A1.

[0010] From WO 2022 / 069258 A1, a device and a method for detecting anomalies in an industrial plant for carrying out a production process are known.

[0011] The invention is based on the objective of further developing a method and a system for evaluating a technical plant.

[0012] The problem is solved by a method for evaluating a technical system with the features specified in claim 1. Advantageous embodiments and further developments are the subject of the dependent claims. The problem is also solved by a system for evaluating a technical system with the features specified in claim 15.

[0013] A method for evaluating a technical system is proposed. During an operational phase, an operational data record is recorded. This operational data record is assigned to a class by a classifier. At least one performance indicator is generated from the operational data record. The generated performance indicator is assigned to the operational data record. The performance indicator assigned to the operational data record is then evaluated.

[0014] The technical system is, for example, a drive system comprising an electric motor, a converter for generating three-phase alternating current for the electric motor, and a gearbox for reducing the motor's speed. The technical system could be, for example, a rotary table, a conveyor belt, or a storage and retrieval machine. Each class is assigned a specific application cycle. Possible application cycles include forward travel, reverse travel, motor operation, generator operation, transport of a light load, and transport of a heavy load. The classifier assigns the recorded operational data record to the class to which the application cycle in which the operational data record was recorded is assigned.

[0015] The recorded operational data set comprises numerous measured variables with many measured values, which are not directly evaluable by a user. By reducing this extensive operational data set to one or a few key performance indicators (KPIs), evaluation becomes possible. Potential KPIs can be defined, for example, by fixed calculation rules such as minimum value calculation, maximum value calculation, average value calculation, or quantile calculation. Alternatively, a KPI can be calculated by a model specifically trained on the reference values, such as a model for calculating anomaly values. The method according to the invention thus allows for the evaluation of the technical system during operation across different application cycles.

[0016] According to an advantageous embodiment of the invention, the recorded operating data set comprises a plurality of measured variables. Each measured variable comprises a plurality of measured values. The measured values ​​of each measured variable are recorded sequentially. For example, the recorded operating data set comprises eight different measured variables, and each measured variable comprises 2048 measured values. For example, the measured values ​​of each measured variable are recorded at equidistant time intervals of 5 ms each. Examples of measured variables are DC link voltage, current, frequency, rotational speed, and torque.

[0017] According to an advantageous embodiment of the invention, a dimensionality reduction of the acquired operational data set is performed before the operational data set is assigned to a class by the classifier. The dimensionality reduction is performed, for example, by principal component analysis. Principal component analysis is a mathematical method for approximating a large number of statistical variables by a smaller number of the most meaningful linear combinations of said variables. Principal component analysis is described, for example, in the document "An Introduction to Statistical Learning," page 374. According to an advantageous embodiment of the invention, the operational data set is acquired when a defined trigger condition is met and when a defined time interval has elapsed since the acquisition of the previous operational data set.

[0018] One such trigger condition is, for example, the start of an application cycle of the technical system when the speed of an electric motor in the system exceeds a predefined minimum speed. This ensures that the recorded operating data record originates from an application cycle and can therefore be assigned to a class.

[0019] For example, the time period is defined as six hours. This means that only four operational data records are captured per day. Therefore, the number of operational data records to be captured and stored is advantageously reduced.

[0020] Experience has shown that the health status of a technical system changes slowly; therefore, the relatively infrequent recording of operational data sets between relatively long periods of time is sufficient.

[0021] According to an advantageous embodiment of the invention, the performance indicator assigned to the operating data set is evaluated by comparing the performance indicator with at least one limit value previously assigned to the class. If the limit value is exceeded, a message is issued.

[0022] According to an advantageous embodiment of the invention, several reference data sets are acquired during a preliminary reference phase. Reference data sets that exhibit a relatively high degree of similarity to one another are identified as belonging to a common cluster. The reference data sets belonging to a common cluster are assigned to a common class. Reference data sets that exhibit a relatively high degree of similarity to one another form clusters in a feature space. Such clusters are referred to as clusters.

[0023] According to an advantageous embodiment of the invention, the clusters are automatically detected by a clustering method, for example, DBSCAN. Alternatively, the clusters are detected by manually assigning them to a class. In this way, data sets are created with which the classifier is trained, which later assigns operational data sets to the classes. Reference data sets with a relatively high degree of similarity to each other originate from similar application cycles. Thus, each class is assigned to a specific application cycle.

[0024] According to an advantageous embodiment of the invention, reference data sets that do not exhibit a sufficiently high degree of similarity to any of the other reference data sets are discarded. Reference data sets lacking similarity to other reference data sets originate from unusual application cycles or malfunctions.

[0025] According to an advantageous embodiment of the invention, each recorded reference data set comprises a plurality of measured variables. Each measured variable comprises a plurality of measured values. The measured values ​​of each measured variable are recorded sequentially. A reference data set is recorded when a defined trigger condition is met. For example, each recorded reference data set comprises eight different measured variables, and each measured variable comprises 2048 measured values. For example, the measured values ​​of each measured variable are recorded at equidistant time intervals of 5 ms each. Measured variables include, for example, DC link voltage, current, frequency, rotational speed, and torque. A trigger condition is, for example, the start of an application cycle of the technical system when, for example, the rotational speed of an electric motor of the technical system exceeds a predetermined minimum speed.The recorded reference data set can therefore be assigned to exactly one application cycle.

[0026] According to an advantageous embodiment of the invention, a dimensionality reduction of the recorded reference data sets is performed before the reference data sets are recognized as belonging to a cluster. The dimensionality reduction is performed, for example, by principal component analysis. Principal component analysis is a mathematical method for approximating a large number of statistical variables by a smaller number of highly informative linear combinations of said variables.

[0027] According to an advantageous embodiment of the invention, at least one performance indicator is generated from each reference data set assigned to a class, and the generated performance indicators are assigned to the class. The performance indicators generated from the extensive reference data sets can later be compared with the performance indicators generated from the operational data sets.

[0028] According to an advantageous embodiment of the invention, at least one model is developed from the generated performance indicators for the classes assigned by the classifier, and the determined model is assigned to the class. The determined model evaluates the performance indicator of an operational data set in order to determine its normality with respect to the reference data set.

[0029] According to an advantageous embodiment of the invention, the model comprises at least one limit value determined from the generated performance indicators. The determined limit values ​​can subsequently be compared with the performance indicators generated from the operational data sets.

[0030] According to an advantageous embodiment of the invention, the classifier is trained to assign future recorded operational data records to previously defined classes. The aim is an error-free assignment of recorded operational data records to previously defined classes.

[0031] According to an advantageous embodiment of the invention, the classifier is designed as a support vector machine. A support vector machine finds a hyperplane in space which separates the classes within the training data as optimally as possible.

[0032] A system according to the invention for evaluating a technical plant comprises a data storage device and a classifier. The system according to the invention is configured to carry out the method according to the invention. The system according to the invention allows for the evaluation of the technical plant during operation in different application cycles.

[0033] The invention is not limited to the combination of features stated in the claims. For those skilled in the art, further meaningful combinations of claims and / or individual claim features and / or features of the description and / or the figures will arise, particularly from the problem statement and / or the problem arising from a comparison with the prior art. The invention will now be explained in more detail with reference to the figures. The invention is not limited to the embodiments shown in the figures. The figures represent the subject matter of the invention only schematically. They show:

[0034] Figure 1: a flowchart of the recording of a reference dataset during a reference phase,

[0035] Figure 2: a flowchart of the processing of recorded reference datasets during the reference phase,

[0036] Figure 3: a schematic representation of an assignment of reference data records to classes and

[0037] Figure 4: a flowchart of the processing of recorded operational data records during an operational phase.

[0038] Figure 1 shows a flowchart of the acquisition of a reference data set 10 in a technical plant during a reference phase. The reference phase is carried out in preparation to define classes and generate performance indicators, which are later needed for the evaluation of the technical plant.

[0039] The technical system in question is a drive system. The drive system comprises an electric motor, a converter for generating a three-phase alternating voltage for the electric motor, and a gearbox for reducing the speed of the electric motor.

[0040] During the reference phase, a reference data set 10 is recorded. The recorded reference data set 10 comprises a plurality of measured quantities, for example, eight different measured quantities. Measured quantities include, for example, DC link voltage, current, frequency, speed, torque, and possibly others.

[0041] Each measured quantity comprises a plurality of measured values, for example, 2048 measured values. The measured values ​​of each measured quantity are recorded sequentially. For example, the measured values ​​of each measured quantity are recorded at equidistant time intervals of 5 ms each. The recorded reference data set 10 thus comprises, for example, eight different measured quantities, each with 2048 measured values. The recorded reference data set 10 is stored in a data storage device 20 in a first step 101.

[0042] During the reference phase, several more reference data sets 10 are recorded. The first step 101 is repeated, and the recorded reference data sets 10 are stored in the data memory 20. Another reference data set 10 is recorded when a defined trigger condition is met. Such a trigger condition is, for example, the start of an application cycle of the technical system when, for instance, the speed of an electric motor in the technical system exceeds a predefined minimum speed.

[0043] Figure 2 shows a flowchart of the processing of recorded reference data sets 10 during the reference phase. In step 102, the previously recorded and stored reference data sets 10 are loaded from the data storage 20.

[0044] In step 103, a dimensional reduction of the recorded reference datasets 10 is first performed by means of a principal component analysis. During the principal component analysis, a transformation matrix is ​​determined and applied to the reference datasets 10.

[0045] Reference records 10 that exhibit a relatively high degree of similarity to each other belong to a common cluster and are recognized as belonging to a common cluster. Reference records 10 that belong to a common cluster and have been recognized as belonging to a common cluster are assigned to a common class.

[0046] The clusters are automatically detected using a clustering method, such as DBSCAN. Alternatively, the clusters are detected by manually assigning them a class. Reference data records 10 that do not exhibit a sufficiently high degree of similarity to any of the other reference data records 10 are discarded.

[0047] Reference data records 10 with a relatively high degree of similarity to each other originate from similar application cycles. Each class is assigned a specific application cycle. Therefore, each recorded reference data record 10 can be assigned to exactly one application cycle. Reference data records 10 that show no similarity to other reference data records 10 originate from unusual application cycles or malfunctions and are therefore discarded.

[0048] In step 104, a classifier is trained using the data set generated in step 103, which contains several reference data sets 10 and the classes assigned to the reference data sets 10.

[0049] The classifier is trained using a machine learning method to assign future recorded operational data records (10) to the previously defined classes. The goal is an error-free assignment of recorded operational data records (10) to the previously defined classes.

[0050] In step 105, a performance indicator is generated from each reference data record 10 contained in a class. The generated performance indicators are assigned to the class to which the respective reference data record 10 is assigned.

[0051] In step 106, a model for the technical system to be evaluated is calculated for each class. The calculated models are then stored in data storage 20 together with the generated performance indicators.

[0052] In step 107, the models determine limit values ​​from the generated performance indicators. These limit values ​​are assigned to the class to which the respective performance indicator belongs. In step 106, the determined limit values ​​are stored in data storage 20.

[0053] Figure 3 shows a schematic representation of the assignment of reference data sets 10 to classes, which takes place in steps 103 and 104. As already mentioned, each reference data set 10 comprises a plurality of measured quantities with a plurality of measured values.

[0054] In sub-step 103a, the dimensional reduction of the recorded data is performed.

[0055] Reference data sets 10 were performed, and the reference data sets 10 are in a

[0056] Feature space is shown. In the representation shown here, the reference data sets 10 are only shown two-dimensionally for simplification. Each of the points shown corresponds to a reference data set 10.

[0057] Reference data sets 10, which exhibit at least a relatively high degree of similarity to each other, belong to a common cluster and are located close to each other in the feature space.

[0058] In substep 103b, the reference data records 10 belonging to a common cluster are identified as belonging to that common cluster. Reference data records 10 that do not exhibit a sufficiently high degree of similarity to any of the other reference data records 10 are also discarded. In the example shown here, two reference data records 10 are therefore discarded, i.e., deleted and thus no longer considered.

[0059] The reference data records 10, which belong to a common cluster, are also assigned to a common class by a clustering procedure or manually.

[0060] In sub-step 104a, the classifier is trained using the reference data sets 10, to which a class was assigned in step 103b. In the example shown here, there are three classes. The reference data sets 10 assigned to the classes are each located within the circles shown, which illustrate the classifier.

[0061] Operational data records 12, which later fall within the area of ​​influence of a circle, are assigned by the classifier of the respective class.

[0062] Figure 4 shows a flowchart of the processing of recorded operational data records 12 in the technical plant during an operational phase. The evaluation of the technical plant is carried out during the operational phase.

[0063] During the operating phase, an operating data record 12 is recorded. The recorded operating data record 12 comprises a plurality of measured variables, for example, eight different measured variables. Measured variables include, for example, DC link voltage, current, frequency, speed, torque, and possibly others.

[0064] Each measurand comprises a plurality of measured values, for example, 2048 measured values. The measured values ​​of each measurand are recorded sequentially. For example, the measured values ​​of each measurand are recorded at equidistant time intervals of 5 ms each.

[0065] The recorded operational data set 12 thus comprises, for example, eight different measured variables, each with 2048 measured values. The recorded operational data set 10 is stored in the data memory 20 in a first step 121. In the first step 121, the recorded operational data set 12 is also fed to the classifier.

[0066] In the first step, 121, a dimensionality reduction of the recorded operational data sets 12 is also performed by means of a principal component analysis. The same principal component analysis is carried out that was previously performed during the reference phase, in step 103, to reduce the dimensionality of the recorded reference data sets 10. In particular, the same transformation matrix that was previously applied to the reference data sets 10 during the reference phase is applied to the operational data sets 12.

[0067] In step 122, the operational data record 12 is assigned by the classifier to one of the previously defined classes. Operational data record 12 is assigned to the class to which it has the greatest similarity.

[0068] In step 123, one or more performance indicators are generated from the operational data set 12 and assigned to the operational data set 12.

[0069] In step 124, the class to which operational data record 12 is assigned is stored in database 20 for operational data record 12. In step 124, the generated performance indicators for operational data record 12 are also stored in database 20.

[0070] In step 125, the performance indicators assigned to operational data record 12 are evaluated. For this purpose, the performance indicators assigned to the classes are loaded from data storage 20. The limit values ​​assigned to the classes are also loaded from data storage 20. The performance indicators assigned to operational data record 12 are then compared with the limit values ​​assigned to the classes. If a limit value is exceeded, a corresponding message is issued in step 126.

[0071] Steps 121 to 126 are executed repeatedly. Specifically, these steps are repeated when a defined trigger condition is met and when a defined period of time has elapsed since the previous operational data record was recorded.

[0072] One such trigger condition is, for example, the start of an application cycle of the technical system when, for example, the speed of an electric motor of the technical system exceeds a predetermined minimum speed.

[0073] The time period is defined, for example, as six hours. Therefore, operational data records 12 are only recorded and processed four times a day.

[0074] Reference symbol list

[0075] 10 Reference data set 12 Operational data set

[0076] 20 data storage devices

[0077] 101-108 steps during the reference phase

[0078] 121-126 steps during the reference phase

Claims

Patent claims:

1. Method for evaluating a technical plant, wherein an operating data record (12) is recorded during an operating phase; the operating data record (12) is assigned to a class by a classifier; at least one performance indicator is generated from the operating data record (12); the generated performance indicator is assigned to the operating data record (12); the performance indicator assigned to the operating data record (12) is evaluated.

2. Method according to one of the preceding claims, characterized in that the recorded operational data set (12) comprises a plurality of measured variables, and that each measured variable comprises a plurality of measured values, and that the measured values ​​of each measured variable are recorded successively.

3. Method according to one of the preceding claims, characterized in that a dimensional reduction of the recorded operational data set (12) is carried out before the operational data set (12) is assigned to a class by the classifier.

4. Method according to one of the preceding claims, characterized in that the operating data record (12) is recorded when a defined trigger condition is met, and when a defined period of time has elapsed since the recording of the previous operating data record (12).

5. Method according to one of the preceding claims, characterized in that the performance indicator assigned to the operating data set (12) is evaluated by comparing the performance indicator with at least one limit value previously assigned to the class; and that a message is issued if the limit value is exceeded.

6. A method according to one of the preceding claims, characterized in that several reference data sets (10) are recorded in preparation during a reference phase; and Reference data records (10) which exhibit a relatively high degree of similarity to each other are identified as belonging to a common cluster; and the reference data records (10) belonging to a common cluster are assigned to a common class.

7. Method of claim 6, characterized in that Reference data sets (10) which do not show a sufficiently high degree of similarity to any of the other reference data sets (10) are discarded.

8. Method according to one of claims 6 to 7, characterized in that each recorded reference data set (10) comprises a plurality of measured quantities, and that each measured quantity comprises a plurality of measured values, and that the measured values ​​of each measured quantity are recorded sequentially, and that a reference data set (10) is recorded when a defined trigger condition is met.

9. Method according to one of claims 6 to 8, characterized in that a dimensionality reduction of the recorded reference data sets (10) is carried out before the reference data sets (10) are recognized as belonging to a cluster.

10. Method according to one of claims 6 to 9, characterized in that at least one performance indicator is generated from each reference data set (10) assigned to a class; and the generated performance indicators are assigned to the class.

11. Method of claim 10, characterized in that at least one model is developed from the generated performance indicators for the classes assigned by the classifier, and that the determined model is assigned to the class.

12. Method of claim 11, characterized in that the model includes at least one limit value determined from the generated performance indicators.

13. Method according to one of claims 6 to 12, characterized in that the classifier is trained to assign future recorded operational data sets (12) to the previously defined classes.

14. Method according to one of claims 6 to 13, characterized in that the classifier is designed as a Support Vector Machine.

15. System for evaluating a technical plant, comprising a data storage device (20) and a classifier, wherein the system is configured to perform the method according to one of the preceding claims.

Images