Procedure and system for the assessment of a technical installation

Abstract

A method for evaluating a technical installation 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 indicator assigned to the operating data set is evaluated. A system according to the invention for evaluating a technical installation 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

The invention relates to a method for evaluating a technical installation, wherein an operating data set is recorded during an operating phase, and the operating data set is assigned to a class by a classifier. The invention also relates to a system for evaluating a technical installation, which is set up to carry out the method according to the invention.Technical installations, for example drive systems, have a multiplicity of technical components, for example an electric motor, a converter for generating a three-phase alternating voltage for the electric motor, and a transmission for transmission of a rotational speed of the electric motor. Such technical installations are, for example, turntables, conveyor belts or storage and retrieval devices. The components of such technical installations can have malfunctions after a longer operating time, which can be attributed to wear which has occurred during operation.It is known to monitor such technical installations by recording and evaluating specific measured values at specific times. If the recorded measured values deviate too much from previously defined setpoint values, an error is assumed in the system and a corresponding message is output to the operator. For example, a current generated by the converter, which exceeds a defined limit value, can be an indication of a stiffness of the transmission due to wear.In a regular operation, the technical installation moves, for example, a light load and a heavy load. In order to detect a difficulty, the current is monitored with a limit value. The limit value must be defined in such a way that it is not exceeded under the heavy load. If now a difficulty arises in the technical installation at a light load, the limit value is not exceeded as a result and the difficulty cannot be detected in this way.DE 10 2008 063 924 A1 discloses a method for detecting faults and assigning them to causes of faults in a hydrostatic system and a corresponding control device.DE 10 2013 100 411 A1 discloses a method and a device for monitoring the state of a technical installation.DE 10 2022 114 579 A1 discloses an industrial plant and a method for plant operation and plant monitoring.WO 2020 / 216452 A1 discloses a method for the state analysis of a technical installation.WO 2022 / 069258 A1 discloses an apparatus and a method for detecting anomalies in an industrial installation for carrying out a production process.The object of the invention is to further develop a method and a system for evaluating a technical installation.The object is achieved by a method for evaluating a technical installation having the features specified in claim 1. Advantageous embodiments and developments are the subject matter of the dependent claims. The object is also achieved by a system for evaluating a technical installation having the features specified in claim 13.A method for evaluating a technical installation is proposed. During an operating phase, an operating data set is recorded when a defined trigger condition is fulfilled, and when a defined time period has elapsed since the recording of the previous operating data set. The trigger condition is the start of an application cycle of the technical installation if, for example, a rotational speed of an electric motor of the technical installation exceeds a predefined minimum rotational speed. It is thus ensured that the recorded operating data set originates from an application cycle and can thus be assigned to a class. 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 successively in time. For example, the recorded operating data set comprises eight different measured variables, and each measured variable comprises 848 measured values. For example, the measured values of each measured variable are recorded in equidistant time segments of 5 ms each.The operational dataset is assigned to a class by a classifier. Each class is assigned a specific application cycle. The classifier assigns the recorded operating data set to that class to which the application cycle in which the operating data set was recorded is assigned.At least one performance characteristic number is generated from the operating data set. The generated performance characteristic is assigned to the operating data set. The performance characteristic value assigned to the operating data set is evaluated.According to the invention, a plurality of reference data sets are recorded preliminarily during a reference phase. Reference data sets which have a relatively high similarity to one another are recognized as belonging to a common cluster. The reference data sets associated with a common cluster are associated with a common class. Reference data sets which have a relatively high similarity to one another form clusters in a feature space. Such clusters are referred to as clusters. The classifier is trained with the reference data sets to which a class has been assigned.The technical installation is, for example, a drive system which comprises an electric motor, a converter for generating a three-phase alternating voltage for the electric motor, and a transmission for converting a rotational speed of the electric motor. The technical installation is, for example, a rotary table, a conveyor belt or a storage and retrieval device. Possible application cycles are, for example, forward travel, reverse travel, motor operation, generator operation, transport of a light load and transport of a heavy load.The recorded operating data set comprises many measured variables with many measured values, which cannot be evaluated directly by a user. By reducing the comprehensive operating data set to one or a few performance indicators, an evaluation is possible. Possible performance characteristics are given, for example, by fixed calculation rules such as minimum value formation, maximum value formation, mean value formation or quantile formation. However, a performance characteristic can also be calculated by a model specifically learned on the reference values, for example a model for calculating anomaly values. The method according to the invention thus allows the technical installation to be evaluated during operation in different application cycles.Measured variables are, for example, intermediate circuit voltage, current, frequency, rotational speed, torque.According to an advantageous embodiment of the invention, a dimensional reduction of the recorded operating data set is carried out before the operating data set is assigned to a class by the classifier. The dimension reduction is performed by, for example, principal component analysis. The principal component analysis is a mathematical method for approximating a plurality of statistical variables by a smaller number of linear combinations of said variables which are as meaningful as possible. The principal component analysis is described, for example, in the document "An Introduction to Statistical Learning, page 374.For example, the time period is defined as six hours. Thus, only four recordings of operating data sets are carried out on a day. Thus, the quantity of the operating data sets to be recorded and stored is advantageously reduced. Experience has shown that a state of health of the technical installation changes slowly, and therefore the relatively rare recordings of operating datasets between relatively large periods of time are sufficient.According to an advantageous embodiment of the invention, the performance characteristic value assigned to the operating data set is evaluated by comparing the performance characteristic value with at least one limit value assigned beforehand to the class. If the limit value is exceeded, a message is output.According to an advantageous embodiment of the invention, the clusters are recognized automatically by a clustering method, for example DBSCAN. Alternatively, the clusters are recognized by manually assigning a class.In this way, data sets are formed with which the classifier is trained, from which subsequently operating data sets are assigned to the classes. Reference data sets with a relatively high similarity to one another originate from application cycles of the same type. Thus, each class is assigned a specific application cycle.According to an advantageous embodiment of the invention, reference data sets which do not have a sufficiently high similarity to any of the other reference data sets are discarded. Reference data sets without similarity to other reference data sets originate from unusual application cycles or accidents.According to an advantageous embodiment of the invention, each recorded reference data set comprises a plurality of measurement variables. Each measured variable comprises a plurality of measured values. The measured values of each measured variable are recorded successively in time. A reference data set is recorded when a defined trigger condition is fulfilled. For example, each recorded reference data set comprises eight different measured variables, and each measured variable comprises 848 measured values. For example, the measured values of each measured variable are recorded in equidistant time segments of 5 ms each. Measured variables are, for example, intermediate circuit voltage, current, frequency, rotational speed, torque. A trigger condition is, for example, the start of an application cycle of the technical installation if, for example, a rotational speed of an electric motor of the technical installation exceeds a predefined minimum rotational speed. The recorded reference data set can thus be assigned to exactly one application cycle.According to an advantageous embodiment of the invention, a dimensional reduction of the recorded reference data sets is carried out before the reference data sets are recognized as belonging to a cluster. The dimension reduction is performed by, for example, principal component analysis. The principal component analysis is a mathematical method for approximating a plurality of statistical variables by a smaller number of linear combinations of said variables which are as meaningful as possible.According to an advantageous embodiment of the invention, at least one performance characteristic number is generated from each reference data set assigned to a class, and the generated performance characteristics numbers are assigned to the class. The performance indicators generated from the comprehensive reference data sets are later comparable to the performance indicators generated from the operating data sets.According to an advantageous embodiment of the invention, at least one model is developed for the classes assigned by the classifier from the generated performance characteristics, and the determined model is assigned to the class. The ascertained model evaluates the performance characteristic of an operating dataset in order to ascertain its normality with respect to the reference dataset.According to an advantageous embodiment of the invention, the model comprises at least one limit value determined from the generated performance characteristics. The limit values determined are later comparable to the performance characteristics generated from the operating data sets.According to an advantageous embodiment of the invention, the classifier is trained in order to assign operating datasets recorded in the future to the previously defined classes. The aim is a fault-free assignment of recorded operating datasets to previously defined classes.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 from one another as optimally as possible within the training data.A system according to the invention for evaluating a technical installation comprises a data memory and a classifier. The system according to the invention is configured to execute the method according to the invention. The system according to the invention allows the technical installation to be evaluated during operation in different application cycles.The invention is not limited to the combination of features of the claims. The skilled person will be familiar with further meaningful combination possibilities of claims and / or individual claim features and / or features of the description and / or of the figures, in particular from the task and / or the task which arises by comparison with the prior art.The invention will now be explained in more detail with reference to figures. The invention is not limited to the exemplary embodiments shown in the figures. The figures only schematically represent the subject matter of the invention. The following are shown: FIG. 1 : a flow diagram of a recording of a reference data set during a reference phase, FIG. 2 : shows a flow diagram of a processing of recorded reference data sets during the reference phase, FIG. 3 : shows a schematic illustration of an assignment of reference data sets to classes, and FIG. 4 : shows a flow diagram of a processing of recorded operating datasets during an operating phase.FIG. 1 shows a flow diagram of a recording of a reference data set 10 in a technical installation during a reference phase. The reference phase is preliminarily carried out in order to define classes and to generate performance characteristics which are later required for evaluating the technical installation.The technical installation is in the present case 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 transmission for converting a rotational speed of the electric motor.During the reference phase, a reference data set 10 is recorded. The recorded reference data set 10 comprises a plurality of measurement variables, for example eight different measurement variables. Measured variables are, for example, intermediate circuit voltage, current, frequency, rotational speed, torque and, if appropriate, further values.Each measured variable comprises a plurality of measured values, for example, 848 measured values. The measured values of each measured variable are recorded successively in time. For example, the measured values of each measured variable are recorded in equidistant time segments of 5 ms each.The recorded reference data set 10 thus comprises, for example, eight different measured variables, each having 848 measured values. The recorded reference data set 10 is stored in a data memory 20 in a first step 101.During the reference phase, a plurality of reference data sets 10 are still recorded. The first step 101 is repeated and the recorded reference data sets 10 are stored in the data memory 20. A further reference data set 10 is recorded when a defined trigger condition is fulfilled. Such a trigger condition is, for example, the start of an application cycle of the technical installation if, for example, a rotational speed of an electric motor of the technical installation exceeds a predefined minimum rotational speed.FIG. 2 shows a flow chart of a processing of recorded reference data sets 10 during the reference phase. In a step 102, the reference data sets 10 previously recorded and stored are loaded from the data memory 20.In a step 103, firstly a dimensional reduction of the recorded reference data sets 10 is carried out by a principal component analysis. When performing the principal component analysis, a transformation matrix is determined and applied to the reference data sets 10.Reference data sets 10 which have a relatively high similarity to one another belong to a common cluster and are recognized as belonging to a common cluster. Reference data sets 10 which belong to a common cluster and have been recognized as belonging to a common cluster are assigned to a common class in this case.The clusters are recognized automatically by a clustering method, for example DBSCAN. Alternatively, the clusters are recognized by manually assigning a class. Reference data sets 10 which do not have a sufficiently high similarity to any of the other reference data sets 10 are discarded.Reference data sets 10 with a relatively high similarity to one another originate from application cycles of the same type. Each class is assigned a specific application cycle. Each recorded reference data set 10 can thus be assigned to exactly one application cycle. Reference data sets 10 without similarity to other reference data sets 10 originate from unusual application cycles or accidents and are therefore discarded.In a step 104, a classifier is trained with the dataset generated in step 103, which contains a plurality of reference datasets 10 and the classes assigned to the reference datasets 10.In this case, the classifier is trained by means of a machine learning method in order to assign operating datasets 10 recorded in the future to the previously defined classes. The aim is a fault-free assignment of recorded operating datasets 10 to previously defined classes.In a step 105, a performance characteristic number is generated from each reference data set 10 contained in a class. The generated performance indicators are assigned to the class to which the respective reference data set 10 is assigned.In a step 106, a model for technical installation to be evaluated is calculated for each class. The calculated models are then stored together with the generated performance characteristics in the data memory 20.In a step 107, limit values are determined from the generated performance characteristics by the models. The limit values determined are assigned to the class to which the respective performance characteristic number is assigned. In a step 106, the determined limit values are stored in the data memory 20.FIG. 3 shows a schematic illustration of an 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 variables with a plurality of measured values.In a substep 103 a, the dimensional reduction of the recorded reference data records 10 is carried out, and the reference data records 10 are displayed in a feature space. In the illustration shown here, the reference data sets 10 are only illustrated two-dimensionally for the sake of simplicity. Each of the points shown corresponds to a reference data set 10.Reference data sets 10, which have at least a relatively high similarity to one another, belong to a common cluster and are close to one another in the feature space.In a substep 103 b, the reference data sets 10 which belong to a common cluster are recognized as belonging to a common cluster. Those reference data sets 10 which do not have a sufficiently high similarity to any of the other reference data sets 10 are also discarded. In the example shown here, two reference data sets 10 are thus discarded, i.e. deleted and thus no longer taken into account.Also, the reference data sets 10 belonging to a common cluster are assigned to a common class by a clustering method or manually.In a substep 104 a, the classifier is trained with the reference data sets 10 to which a class has been assigned in step 103 b. In the example illustrated here, there are three classes. The reference data sets 10 assigned to the classes are each located within the circles shown, which visualize the classifier by way of example.Operational datasets 12 that later fall within the influence space of a circle are assigned to the respective class by the classifier.FIG. 4 shows a flow diagram of a processing of recorded operating datasets 12 in the technical installation during an operating phase. In the operating phase, the evaluation of the technical installation is carried out.During the operating phase, an operating data set 12 is recorded. The recorded operating data set 12 comprises a plurality of measured variables, for example eight different measured variables. Measured variables are, for example, intermediate circuit voltage, current, frequency, rotational speed, torque and, if appropriate, further values.Each measured variable comprises a plurality of measured values, for example, 848 measured values. The measured values of each measured variable are recorded successively in time. For example, the measured values of each measured variable are recorded in equidistant time segments of 5 ms each.The recorded operating data set 12 thus comprises, for example, eight different measured variables, each having 848 measured values. The recorded operating data record 10 is stored in the data memory 20 in a first step 121. In the first step 121, the recorded operating data set 12 is also supplied to the classifier.In the first step 121, a dimensional reduction of the recorded operating datasets 12 is also carried out by a principal component analysis. In this case, the same main component analysis is carried out which was previously carried out during the reference phase, in step 103, for the dimensional reduction of the recorded reference data sets 10. In particular, the same transformation matrix is applied to the operating datasets 12 as was previously applied to the reference datasets 10 during the reference phase.In a step 122, the operating data set 12 is assigned by the classifier to one of the previously defined classes. The operating data set 12 is assigned to that class to which the greatest similarity exists.In a step 123, one or more performance indicators are generated from the operating data record 12 and are assigned to the operating data record 12.In a step 124, the class to which the operating data record 12 is assigned is stored in the database 20 with respect to the operating data record 12. In step 124, the generated performance characteristics relating to the operating dataset 12 are also stored in the database 20.In a step 125, the performance indicators assigned to the operating data set 12 are evaluated. For this purpose, the performance indicators assigned to the classes are loaded from the data memory 20. The limit values assigned to the classes are also loaded from the data memory 20. The performance indicators assigned to the operating data set 12 are then compared with the limit values assigned to the classes.If a limit value is exceeded, a corresponding message is output in a step 126.The above steps 121 to 126 are repeatedly executed. In particular, the steps are repeated if a defined trigger condition is fulfilled and if a defined time period has elapsed since the recording of the previous operating data set.Such a trigger condition is, for example, the start of an application cycle of the technical installation if, for example, a rotational speed of an electric motor of the technical installation exceeds a predefined minimum rotational speed.The time period is defined as six hours, for example. Operating datasets 12 are therefore recorded and processed only four times a day.List of reference characters10 Reference data set 12 Operating data set 20 Data memory 101..108 Steps during the reference phase 121..126 Steps during the reference phase

Claims

Method for evaluating a technical installation, wherein an operating data set (12) is recorded during an operating phase when a defined trigger condition is fulfilled, and when a defined period of time has elapsed since the recording of the previous operating data set (12), wherein the trigger condition is the start of an application cycle of the technical installation, and wherein the recorded operating data set (12) comprises a plurality of measurement variables, and wherein each measurement variable comprises a plurality of measurement values, and wherein the measurement values of each measurement variable are recorded successively in time; the operating data set (12) is assigned to a class by a classifier, wherein a specific application cycle of the technical installation is assigned to each class, and wherein the classifier assigns the recorded operating data set to the class which is assigned to the application cycle in which the operating data set was recorded; at least one performance characteristic number is generated from the operating data set (12); the generated performance characteristic is assigned to the operating data set (12); the performance characteristic assigned to the operating data set (12) is evaluated, and wherein a plurality of reference data sets (10) are recorded preliminarily during a reference phase; and reference data sets (10), which have a relatively high similarity to one another, are recognized as belonging to a common cluster; and the reference data sets (10) belonging to a common cluster are assigned to a common class, and wherein the classifier is trained with the reference data sets (10) to which a class has been assigned.Method according to one of the preceding claims, characterized in that the measured variables are an intermediate circuit voltage and / or a current and / or a frequency and / or a rotational speed and / or a torque.Method according to one of the preceding claims, characterized in that a dimensional reduction of the recorded operating data set (12) is carried out before the operating data set (12) is assigned to a class by the classifier.Method according to one of the preceding claims, characterized in that the performance characteristic number assigned to the operating data record (12) is evaluated by comparing the performance characteristic number with at least one limit value assigned beforehand to the class; and in that a message is output if the limit value is exceeded.Method according to one of the preceding claims, characterized in that reference data sets (10) which do not have a sufficiently high similarity to any of the other reference data sets (10) are discarded.Method according to one of the preceding claims, characterized in that each recorded reference data set (10) comprises a plurality of measurement variables, and in that each measurement variable comprises a plurality of measurement values, and in that the measurement values of each measurement variable are recorded successively in time, and in that a reference data set (10) is recorded if a defined trigger condition is fulfilled.Method according to one of the preceding claims, characterized in that a dimensional reduction of the recorded reference data records (10) is carried out before the reference data records (10) are recognized as belonging to a cluster.Method according to one of the preceding claims, characterized in that at least one performance characteristic number is generated from each reference data set (10) assigned to a class; and the generated performance characteristics numbers are assigned to the class.Method according to Claim 8, characterized in that at least one model is developed from the generated performance characteristics for the classes assigned by the classifier, and in that the ascertained model is assigned to the class.Method according to Claim 9, characterized in that the model comprises at least one limit value determined from the generated performance characteristics.Method according to one of the preceding claims, characterized in that the classifier is trained in order to assign operating datasets (12) recorded in the future to the previously defined classes.Method according to one of the preceding claims, characterized in that the classifier is designed as a support vector machine.A system for evaluating a technical installation, comprising a data memory (20) and a classifier, the system being configured to carry out the method according to any one of the preceding claims.

Images