Industrial equipment fault monitoring model construction method and device, equipment and medium

Abstract

The invention discloses an industrial equipment fault monitoring model construction method and device, equipment and a medium, and relates to the field of artificial intelligence, and the method comprises the steps: obtaining a to-be-processed data set from a target database to construct a target training data set and a target test data set; the target database is a database used for recording industrial equipment fault data; inputting the target training data set into the target fault monitoring model for training, inputting the target test data set into the trained target fault monitoring model, and judging whether an output result meets a model release condition or not; if the model release condition is not met and the preset model structure optimization condition is not met, skipping to the step of obtaining the to-be-processed data set from the target database; and if the preset model structure optimization condition is met, executing a model structure optimization operation, and skipping to the step of obtaining the to-be-processed data set from the target database. According to the method and the device, the fault monitoring precision of the industrial equipment is improved, and the universality and the logic robustness are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a method, device, equipment and medium for constructing an industrial equipment fault monitoring model. Background Art

[0002] Currently, when monitoring the status of industrial equipment during use, we mainly rely on fixed thresholds and static rules to determine the possibility of failure of industrial equipment within a certain period of time. This leads to low monitoring accuracy and the impact of small changes on industrial equipment is easily ignored. It does not take into account the natural aging trend of equipment over time, and is unable to autonomously learn and optimize according to changes in application scenarios and data trends. It lacks universality and logical robustness.

[0003] In summary, how to improve the accuracy of fault monitoring of industrial equipment and enhance its universality and logical robustness is a problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, equipment and medium for constructing an industrial equipment fault monitoring model, which can improve the accuracy of fault monitoring of industrial equipment and enhance universality and logical robustness. The specific scheme is as follows:

[0005] In a first aspect, the present application discloses a method for constructing an industrial equipment fault monitoring model, comprising:

[0006] A target data set reading operator is used to obtain a data set to be processed from a target database, a target data set processing operator is used to process the data set to be processed to obtain target data, and a target data set construction operator is used to construct a target training data set and a target test data set based on the target data; the target database is a database for recording industrial equipment fault data;

[0007] Inputting the target training data set into a target fault monitoring model to train the target fault monitoring model, inputting the target test data set into a new target fault monitoring model obtained through training, and determining whether an output result of the target fault monitoring model meets a model release condition;

[0008] If the output result does not meet the model publishing condition, determine whether the output result meets the preset model structure optimization condition, and when the output result does not meet the preset model structure optimization condition, perform a model iteration operation to jump to the step of obtaining the to-be-processed data set from the target database using the target data set reading operator;

[0009] If the output result satisfies the preset model structure optimization condition, a model structure optimization operation is performed to optimize the model structure of the target fault monitoring model, and the process jumps to the step of obtaining the data set to be processed from the target database using the target data set reading operator;

[0010] Among them, the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visualization modeling platform, and the target fault monitoring model is a model constructed through the target visualization modeling platform.

[0011] Optionally, before using the target data set reading operator to obtain the data set to be processed from the target database, the method further includes:

[0012] The industrial equipment fault data is stored using a pre-created target data set object, and the target data set object is uploaded to the target database in the form of an offline data file.

[0013] Optionally, the using a target data processing operator to process the data set to be processed to obtain target data includes:

[0014] The target data standardization operator is used to standardize the data set to be processed to obtain standardized data, and the target data cleaning operator is used to clean the standardized data to obtain the target data; the target data standardization operator includes a string to date operator and a column filtering operator, and the target data cleaning operator includes a timestamp alignment operator and a missing value processing operator.

[0015] Optionally, the using the target data set construction operator to construct a target training data set and a target test data set based on the target data includes:

[0016] Determine matching data corresponding to each column of data in the target data, and compare each column of data with the corresponding matching data using a column aggregation operator to remove noise data in the target data;

[0017] The target data after noise data is removed is partitioned using a data partitioning operator to construct the target training data set and the target test data set.

[0018] Optionally, the model publishing condition is that the accuracy of the output result is greater than or equal to a first accuracy threshold, and the preset model structure optimization condition is that the accuracy of the output result is less than a second accuracy threshold, and the second accuracy threshold is less than the first accuracy threshold.

[0019] Optionally, the industrial equipment fault monitoring model construction method further includes:

[0020] If the output result meets the model publishing conditions, the target fault monitoring model is uploaded to the target model management library, so that the target fault monitoring model can be downloaded from the target model management library through the model edge reasoning end and deployed locally on the model edge reasoning end.

[0021] Optionally, the industrial equipment fault monitoring model construction method further includes:

[0022] Obtain feedback information uploaded to the target database by the model edge inference end, optimize the target fault monitoring model based on the feedback information to obtain a new target fault monitoring model, and jump to the step of uploading the target fault monitoring model to the target model management library.

[0023] In a second aspect, the present application discloses a device for constructing an industrial equipment fault monitoring model, comprising:

[0024] A data set construction module, used to obtain a data set to be processed from a target database using a target data set reading operator, process the data set to be processed using a target data processing operator to obtain target data, and construct a target training data set and a target test data set based on the target data using a target data set construction operator; the target database is a database for recording industrial equipment fault data;

[0025] A model training module is used to input the target training data set into a target fault monitoring model to train the target fault monitoring model, input the target test data set into a new target fault monitoring model obtained through training, and determine whether an output result of the target fault monitoring model meets a model release condition;

[0026] A model iteration module, configured to determine whether the output result satisfies a preset model structure optimization condition if the output result does not satisfy the model publishing condition, and to perform a model iteration operation to jump to the step of obtaining a to-be-processed data set from a target database using a target data set reading operator when the output result does not satisfy the preset model structure optimization condition;

[0027] A model optimization module, configured to perform a model structure optimization operation to optimize the model structure of the target fault monitoring model if the output result satisfies the preset model structure optimization condition, and jump to the step of obtaining a to-be-processed data set from a target database using a target data set reading operator;

[0028] Among them, the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visualization modeling platform, and the target fault monitoring model is a model constructed through the target visualization modeling platform.

[0029] In a third aspect, the present application discloses an electronic device, comprising:

[0030] Memory, used to store computer programs;

[0031] The processor is used to execute the computer program to implement the aforementioned industrial equipment fault monitoring model construction method.

[0032] In a fourth aspect, the present application discloses a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned industrial equipment fault monitoring model construction method.

[0033] In the present application, when constructing an industrial equipment fault monitoring model, a target data set reading operator is used to obtain a data set to be processed from a target database, a target data processing operator is used to process the data set to be processed to obtain target data, and a target data set construction operator is used to construct a target training data set and a target test data set based on the target data; the target database is a database for recording industrial equipment fault data; the target training data set is input into a target fault monitoring model to train the target fault monitoring model, the target test data set is input into a new target fault monitoring model obtained through training, and it is determined whether the output result of the target fault monitoring model meets the model release condition; if the output result does not meet the model release condition, it is determined that The output result satisfies the preset model structure optimization condition, and when the output result does not meet the preset model structure optimization condition, a model iteration operation is performed to jump to the step of obtaining the data set to be processed from the target database by using the target data set reading operator; if the output result meets the preset model structure optimization condition, a model structure optimization operation is performed to optimize the model structure of the target fault monitoring model, and jump to the step of obtaining the data set to be processed from the target database by using the target data set reading operator; wherein the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visual modeling platform, and the target fault monitoring model is a model constructed through the target visual modeling platform.

[0034] It can be seen that the present application realizes the design of the fault monitoring process through the target visualization modeling platform, and uses various operators to analyze and process the data obtained from the target database. Since the data sets to be processed stored in the target database are all fault data, the processed fault data are used to construct the target training data set and the target test data set, and then the target training data set is used to train the target fault monitoring model to obtain a target fault monitoring model with fault monitoring capability, and the target test data set is input into the target fault monitoring model. The target fault monitoring model will judge the possibility of failure of the industrial equipment corresponding to each test data in the target test data set and output it. If the output result does not meet the model release conditions and does not meet the preset model structure optimization conditions, the data set to be processed is obtained from the target database again for iterative training to improve the fault monitoring capability of the target fault monitoring model, that is, the accuracy of the output result; if the output result meets the preset model structure optimization conditions, it means that the model structure needs to be adjusted, or the relevant parameters need to be optimized, then the model structure of the target fault monitoring model is adjusted, and then jump to the target data set reading operator to obtain the data set to be processed from the target database to retrain the target fault monitoring model. Through multiple iterations and structural optimization of the target fault monitoring model, the accuracy of the target fault monitoring model in fault monitoring of industrial equipment based on the operating data of the industrial equipment is improved, and the universality and logical robustness are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0036] Figure 1 A flow chart of a method for constructing an industrial equipment fault monitoring model disclosed in this application;

[0037] Figure 2 A schematic diagram of a data access process disclosed in this application;

[0038] Figure 3 A schematic diagram of a specific industrial equipment fault monitoring model construction method disclosed in this application;

[0039] Figure 4 A schematic diagram of a data standardization processing flow disclosed in this application;

[0040] Figure 5 A schematic diagram of a data cleaning process disclosed in this application;

[0041] Figure 6 A schematic diagram of a model input set construction process disclosed in this application;

[0042] Figure 7 A schematic diagram of a model construction process disclosed in this application;

[0043] Figure 8 A schematic diagram of a model verification process disclosed in this application;

[0044] Fig. 9 A schematic diagram of a model evaluation process disclosed in this application;

[0045] Fig.10 A schematic diagram of a model data distribution process disclosed in this application;

[0046] Fig.11 A schematic diagram of a device structure for building an industrial equipment fault monitoring model disclosed in this application;

[0047] Fig.12 This is a structural diagram of an electronic device disclosed in this application. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] At present, when monitoring the status of industrial equipment during use, fixed thresholds and static rules are mainly relied on to judge the possibility of failure of industrial equipment within a certain period of time, resulting in low monitoring accuracy, and the impact of small changes on industrial equipment is easily ignored. The natural aging trend of equipment over time is not taken into account, and it is impossible to autonomously learn and optimize according to changes in application scenarios and data trends. It lacks universality and logical robustness. In order to solve the above technical problems, this application discloses a method for constructing an industrial equipment fault monitoring model, which can improve the accuracy of fault monitoring of industrial equipment and enhance universality and logical robustness.

[0050] See also Figure 1 As shown, an embodiment of the present invention discloses a method for constructing an industrial equipment fault monitoring model, including:

[0051] Step S11, using the target data set reading operator to obtain the data set to be processed from the target database, using the target data processing operator to process the data set to be processed to obtain target data, and using the target data set construction operator to construct a target training data set and a target test data set based on the target data; the target database is a database for recording industrial equipment fault data.

[0052] In this embodiment, before using the target data set reading operator to obtain the data set to be processed from the target database, data access is required, that is, to prepare available data for model training and model prediction, such as Figure 2 As shown, the data access process may specifically include: using a pre-created target data set object to store industrial equipment fault data, and uploading the target data set object to the target database in the form of an offline data file. In addition, industrial equipment fault data can also be obtained from a real-time data acquisition system or a business database for model reasoning. In this embodiment, the training sample set is obtained by uploading a data file to the target database, and the real data is obtained by real-time collection through docking with the Internet of Things platform system. After the data is successfully obtained, it will be stored in the target database, and the functional layer will manage the data through the data set management function. Therefore, when iteratively training the target fault monitoring model, you can choose to re-acquire a new data set to be processed from the target database.

[0053] In this embodiment, the target dataset reading operator can be used to obtain the data to be processed from the target database. Specifically, in the target visual modeling platform, drag and drop the dataset reading operator in the operator selector, configure the operating parameters of the operator, and then execute it to read, load and use the corresponding sample dataset (that is, the dataset to be processed). Figure 3 As shown in the figure, there are multiple operators in the operator selector of the target visual modeling platform. Users can select the corresponding operator according to their own needs to realize data analysis and processing. The operator selector provides a rich, general, and robust operator library with high reusability.

[0054] In this embodiment, in order to ensure that the data structure can meet the model usage requirements and improve the model convergence, it is necessary to use the target data processing operator to perform corresponding processing on the acquired data set to be processed, which may include: using the target data standardization operator to standardize the data set to be processed to obtain standardized data, and using the target data cleaning operator to clean the standardized data to obtain the target data; the target data standardization operator includes a string to date operator and a column filtering operator, and the target data cleaning operator includes a timestamp alignment operator and a missing value processing operator. In other words, Figure 4As shown, users can select appropriate data processing operators in the operator selector of the visual modeling workbench according to business standardization requirements to build a data standardization process and complete the corresponding configuration. Taking the implementation of "kiln heating rod fault monitoring" as an example, in this step, the string to date operator and column filter operator are selected to implement the data standardization process. The "date timestamp" column in the sample data set is converted from a string type to a date type through the string to date operator to meet the inspection, processing, and use of date timestamp type data in subsequent processes. Use the column filter operator to filter out unnecessary columns in the sample data and retain only the required data columns to obtain standardized data.

[0055] In this embodiment, after obtaining the standardized data, it is necessary to perform data cleaning on the standardized data to ensure the accuracy, consistency and reliability of the data, help eliminate noise and redundancy in the data, and improve the efficiency of data analysis and modeling. Figure 5 As shown, in order to realize the kiln heating rod fault monitoring, the timestamp alignment operator and the missing value processing operator can be selected to realize the data cleaning process. In this process, the timestamp alignment operator is first used to check the consistency of sampling of the timestamp column of the standardized data within the selected time range, and the missing sampling time value is inserted to fill the missing time value at the "skip" check (time value is missing or invalid) to ensure the integrity of the data. Then the "temperature value" column of the standardized data is checked and processed by the missing value processing operator to make up for the shortcomings of the standardized data in data integrity and improve the data quality. Specifically, the missing value processing operator configured in this embodiment is to process missing values ​​by mean interpolation. This method will take the average of the previous and next non-missing values ​​and insert it into the missing value position. In addition, the user can choose to use the missing value processing operator to fill the missing values ​​of the standardized data with other missing value processing methods such as value filling, linear interpolation, maximum value, minimum value, median, intermediate mean, fixed constant, next value, last value, and row removal. The cleaned and standardized data is easier for the model to understand and process, which helps to discover patterns and trends in the data and enhance the accuracy of business analysis decisions of the subsequently constructed industrial equipment fault monitoring model.

[0056] In this embodiment, Figure 6As shown, after obtaining the target data set, the target data set construction operator can be used to construct the target training data set and the target test data set based on the target data. Specifically, it may include: determining the matching data corresponding to each column of data in the target data, and using the column aggregation operator to compare each column of data with the corresponding matching data to remove the noise data in the target data; using the data partitioning operator to partition the target data after removing the noise data to construct the target training data set and the target test data set. That is to say, in the operator selector of the visual modeling workbench, select the appropriate data construction operator according to the data situation and adjust the configuration parameters to realize the input set construction process. Taking the realization of "kiln heating rod fault monitoring" as an example, in this step, the column lag operator, column aggregation operator and data partitioning operator are selected to realize the model input set construction. When determining the matching data corresponding to each column of data in the target data, the column lag operator can be used to set the same period of data in the target data set to perform sequence matching with the previous period, next period, early period and late period data. Then, the data is compared through the column aggregation operator. If the data values ​​of the comparison column data are all within a reasonable trend range, no operation is performed. If more obvious noise data is screened out, the data is discarded to reduce the impact of dirty data on model training. Finally, the sample data is divided into two groups of data using the data partitioning operator, namely the target training data set and the target test data set. In the case of the present invention, the selected data partitioning operator is configured to perform data partitioning in "relative partitioning", that is, a percentage is specified to divide the sample data into two groups according to the proportion (the setting value is 80, i.e. 80 / 20 partitioning). In addition, data partitioning operators with data partitioning methods such as "absolute partitioning", "top sampling", "linear sampling", "random sampling", and "stratified sampling" are also supported to partition data.

[0057] Step S12: input the target training data set into the target fault monitoring model to train the target fault monitoring model, input the target test data set into the trained new target fault monitoring model, and determine whether the output result of the target fault monitoring model meets the model release conditions.

[0058] In this embodiment, after the target data set is processed into a target training data set and a target test data set, the target training data set can be used to train the target fault monitoring model, wherein the target fault monitoring model is a model constructed by combining the target visualization modeling platform with the operator selector according to business needs. Figure 7As shown in the figure, taking the implementation of "kiln heating rod fault monitoring" as an example, if Keras (an open source artificial neural network library) is used, the Keras Input Layer operator, Keras LSTM Layer operator, Keras Dense Layer operator and Keras Network Learner operator are selected to realize the construction of the kiln heating rod fault monitoring model, and after the construction is completed, the target training data set is input to the input layer Keras NetworkLearner of the model to realize model training. The LSTM (Long Short-Term Memory) algorithm selected and used above is the best practice solution for the kiln heating rod fault monitoring model, but its implementation and construction method design includes but is not limited to the use of other RNN (RNN Recurrent Neural Network) variant algorithm models, CNN (Convolutional Neural Networks) and RNN hybrid models, attention mechanism models, other time series prediction models, etc., which can handle similar problems or achieve the same results, such as Bi-RNN (bidirectional recurrent neural network), Deep-RNN (deep recurrent neural network) model, Transformer model, ARIMA (Autoregressive Integrated Moving Average Model), Prophet model (an open source time series prediction model), CNN model, ConvLSTM (Convolutional Long Short-Term Memory) model and other algorithm models. The LSTM neural network operator group provided by the operator selector of the target visualization modeling platform can effectively process and analyze time series data, and can be combined with deep learning to predict fault trends, providing more accurate and timely alarms than traditional methods.

[0059] After completing the training of the target fault monitoring model, the target test data set can be input into the new target fault monitoring model obtained by training, and the target fault monitoring model can be judged whether it meets the model release condition according to the corresponding output result. The model release condition is that the accuracy of the output result is greater than or equal to the first accuracy threshold. The output result of the target fault monitoring model for the target test data set is the corresponding industrial equipment failure probability, such as "the probability of failure of the kiln heating rod in the next day is 30%". If the fault probability threshold in the output result is set to 50%, it can be determined that the equipment will fail if the failure probability is greater than or equal to 50%, and the equipment will not fail if it is less than 50%. However, both the target training data set and the target test data set are industrial equipment failure data obtained from the target database, so the output results corresponding to the target test data set should theoretically indicate that the equipment will fail. Assuming that the first accuracy threshold is 80%, and there are 100 data in the target test data set, if the corresponding output result has a failure probability of no less than 50% for no less than 80 data, that is, there are more than 80 data corresponding to the output results indicating that the equipment will fail, then it can be determined that the target fault monitoring model has met the model release condition.

[0060] like Figure 8 As shown, in a specific implementation, in this step, the Deployment Loop operator, the numerical scorer operator and the line chart operator are selected to implement the test set result input and comparison verification process. In this process, the loop node of the Deployment Loop operator is used to cyclically deploy the training model and perform data iteration and reuse. The output parameters include the number of loops, loop status, error information and output data set, where the output data set contains the results after the current iteration processing for subsequent analysis or processing.

[0061] It should be pointed out that in actual applications, the fault probability threshold and the first accuracy threshold can be adjusted as needed, such as adjusting the fault probability threshold to 60% and the first accuracy threshold to 85%, so as to improve the prediction accuracy requirements of the target fault monitoring model. In addition, it can also be set that in actual applications, if the probability of equipment failure is greater than or equal to the fault probability threshold, the corresponding fault handling instructions are executed, such as sending alarm information, recording logs, or remotely controlling the equipment to stop running.

[0062] Step S13: If the output result does not meet the model publishing condition, determine whether the output result meets the preset model structure optimization condition, and when the output result does not meet the preset model structure optimization condition, perform a model iteration operation to jump to the step of obtaining the data set to be processed from the target database using the target data set reading operator.

[0063] In this embodiment, the preset model structure optimization condition is that the accuracy of the output result is less than the second accuracy threshold, and the second accuracy threshold is less than the first accuracy threshold. That is to say, assuming that the first accuracy threshold is 80%, the second accuracy threshold is 50%, and the data in the target test data set is 100, if the corresponding output result has a failure probability greater than or equal to 50%, that is, the output result corresponding to more than 50 data indicates that the device will fail, then it can be determined that the output result does not meet the preset model structure optimization condition, and the structure of the current target fault monitoring model can be used for the construction of the industrial equipment fault monitoring model. At this time, it is only necessary to perform the model iteration operation to jump to the step of using the target data set reading operator to obtain the data set to be processed from the target database, so as to optimize the target fault monitoring model. In addition, adjustment strategies such as data optimization, model structure adjustment, parameter adjustment, hyperparameter tuning, and algorithm selection adjustment can also be used to optimize the target fault monitoring model by increasing and / or reducing and / or replacing process nodes, adjusting process node parameter item values, and modifying the business analysis process design.

[0064] It should be pointed out that if Figure 3 As shown, all operators in the embodiment of the present invention can be called by dragging and dropping in the target visualization modeling platform, that is, the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visualization modeling platform. When constructing and optimizing the target fault monitoring model, the user can also design and adjust the fault warning process, steps, and rules through a graphical operation interface. The whole process is intuitive and efficient, and the model parameters can also be dynamically adjusted online according to the online real-time calculation results of the model, which reduces the requirements for programming ability, improves operational flexibility and efficiency, and shortens the model development cycle.

[0065] Step S14: If the output result satisfies the preset model structure optimization condition, a model structure optimization operation is performed to optimize the model structure of the target fault monitoring model, and the process jumps to the step of obtaining the data set to be processed from the target database using the target data set reading operator.

[0066] In this embodiment, if the output result satisfies the preset model structure optimization conditions, that is, assuming that the first accuracy threshold is 80%, the second accuracy threshold is 50%, and there are 100 data in the target test data set, if the corresponding output results have a failure probability of more than 50 data less than 50%, that is, there are more than 50 data corresponding to the output results indicating that the device will not fail, then it can be determined that the output result satisfies the preset model structure optimization conditions, and the current structure of the target fault monitoring model is not suitable for industrial equipment fault monitoring. The model structure needs to be optimized so that the target fault monitoring model can be rebuilt based on the optimized structure, and the step of using the target data set reading operator to obtain the data set to be processed from the target database is jumped to retrain the target fault monitoring model.

[0067] like Fig. 9 As shown in the figure, when the target fault monitoring model has met the model release conditions, the numerical scorer and line graph can be used to analyze the data set output by the loop node, evaluate the result quality of each loop and analyze the data trend to determine whether the current model training has reached the loop end standard. The numerical calculation scorer inputs the predicted value and the actual value, and outputs the mean square error, root mean square error, mean absolute error, etc. to verify and evaluate the accuracy of the prediction results in each loop. The line graph can visualize the predicted value and the actual value of the prediction result, so as to analyze the trend of the predicted value and the actual value during the loop process to observe and analyze the prediction of the model. For the implementation of this step, many general evaluation algorithms are provided in the algorithm selector, including but not limited to cross-validation, confusion matrix, ROC (Receiver Operating Characteristic Curve, Receiver Operating Characteristic Curve), AUC (Area Under Curve, the area under the ROC curve and the coordinate axis), performance measurement, mathematical calculation, and visualization chart method. In the implementation of the kiln heating rod fault monitoring model, when the model reaches the end of training, verification, and tuning, the Regression Evaluator evaluation node is used to evaluate the model overall, and outputs comprehensive performance evaluation indicators including MSE (Mean Squared Error), RMSE (Root MeanSquared Error), MAE (Mean Absolute Error), and R-squared (a coefficient of determination). At the same time, the Regression Evaluator node also supports visual analysis functions, such as residual analysis charts and comparative analysis charts, which can provide a more comprehensive and in-depth analysis and evaluation basis for the overall performance and quality of the model, and the evaluation results can also be used as a reference to determine whether the current target fault monitoring model meets the model release conditions.

[0068] In this embodiment, if the output result meets the model publishing conditions, the target fault monitoring model can be uploaded to the target model management library. All models on the shelf are visible at the model edge reasoning end, and support seamless integration with other data acquisition systems and industrial software. It has good scalability and can adapt to the needs of different industrial scenarios. The model edge reasoning end here is the operating environment for online prediction after the model is launched. Then, at the model edge reasoning end, the target fault monitoring model can be deployed or updated from the target model management library to the model edge reasoning end through the algorithm library. After configuring the data connection address, result push address, and model call task, the model can be successfully started and run at the model edge reasoning end. The process of online prediction after model deployment is roughly divided into three links, namely model data input (access to online real-time data), model call execution, and model result output.

[0069] If the deployment or update is complete, it means that the target fault monitoring model has been officially launched and used at the edge inference end of the model. At this time, the real online data can be used as the model input set to access the target fault monitoring model for monitoring, analysis or prediction. At this time, the data output by the model operation is the result set. If the application side needs it, the result data can be opened through data distribution. Fig.10 As shown in the figure, the result set is distributed through the MQTT (Message Queuing Telemetry Transport) service to support data visualization on the application side. In addition, when the feedback interface of the model edge inference end is opened, the feedback information uploaded to the target database by the model edge inference end can be obtained, so that the target fault monitoring model can be optimized based on the feedback information to obtain a new target fault monitoring model, and the optimized target fault monitoring model can be uploaded to the target model management library, so that the fault detection model can be seamlessly connected to the big data platform, Internet of Things platform, edge computing platform and other multi-resource platform services, which not only solves the problem of model data source limitations, but also greatly expands the computing power resources of the model, so that the model can play a better performance, and at the same time can support the model to realize self-learning and adaptive optimization mechanism, and through continuous data accumulation, the universality and robustness of the model can be continuously enhanced.

[0070] It can be seen that the present application realizes the design of the fault monitoring process through the target visualization modeling platform, and uses various operators to analyze and process the data obtained from the target database. Since the data sets to be processed stored in the target database are all fault data, the processed fault data are used to construct the target training data set and the target test data set, and then the target training data set is used to train the target fault monitoring model to obtain a target fault monitoring model with fault monitoring capability, and the target test data set is input into the target fault monitoring model. The target fault monitoring model will judge the possibility of failure of the industrial equipment corresponding to each test data in the target test data set and output it. If the output result does not meet the model release conditions and does not meet the preset model structure optimization conditions, the data set to be processed is obtained from the target database again for iterative training to improve the fault monitoring capability of the target fault monitoring model, that is, the accuracy of the output result; if the output result meets the preset model structure optimization conditions, it means that the model structure needs to be adjusted, or the relevant parameters need to be optimized, then the model structure of the target fault monitoring model is adjusted, and then jump to the target data set reading operator to obtain the data set to be processed from the target database to retrain the target fault monitoring model. Through multiple iterations and structural optimization of the target fault monitoring model, the accuracy of the target fault monitoring model in fault monitoring of industrial equipment based on the operating data of the industrial equipment is improved, and the universality and logical robustness are enhanced.

[0071] See also Fig.11 As shown, the present application discloses a device for constructing an industrial equipment fault monitoring model, comprising:

[0072] The data set construction module 11 is used to obtain a data set to be processed from a target database using a target data set reading operator, process the data set to be processed using a target data processing operator to obtain target data, and construct a target training data set and a target test data set based on the target data using a target data set construction operator; the target database is a database for recording industrial equipment fault data;

[0073] A model training module 12 is used to input the target training data set into a target fault monitoring model to train the target fault monitoring model, input the target test data set into a new target fault monitoring model obtained through training, and determine whether an output result of the target fault monitoring model meets a model release condition;

[0074] A model iteration module 13 is used to determine whether the output result meets the preset model structure optimization condition if the output result does not meet the model publishing condition, and when the output result does not meet the preset model structure optimization condition, perform a model iteration operation to jump to the step of obtaining the to-be-processed data set from the target database using the target data set reading operator;

[0075] A model optimization module 14 is used for performing a model structure optimization operation to optimize the model structure of the target fault monitoring model if the output result satisfies the preset model structure optimization condition, and jumping to the step of obtaining the data set to be processed from the target database by using the target data set reading operator;

[0076] Among them, the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visualization modeling platform, and the target fault monitoring model is a model constructed through the target visualization modeling platform.

[0077] It can be seen that the present application realizes the design of the fault monitoring process through the target visualization modeling platform, and uses various operators to analyze and process the data obtained from the target database. Since the data sets to be processed stored in the target database are all fault data, the processed fault data are used to construct the target training data set and the target test data set, and then the target training data set is used to train the target fault monitoring model to obtain a target fault monitoring model with fault monitoring capability, and the target test data set is input into the target fault monitoring model. The target fault monitoring model will judge the possibility of failure of the industrial equipment corresponding to each test data in the target test data set and output it. If the output result does not meet the model release conditions and does not meet the preset model structure optimization conditions, the data set to be processed is obtained from the target database again for iterative training to improve the fault monitoring capability of the target fault monitoring model, that is, the accuracy of the output result; if the output result meets the preset model structure optimization conditions, it means that the model structure needs to be adjusted, or the relevant parameters need to be optimized, then the model structure of the target fault monitoring model is adjusted, and then jump to the target data set reading operator to obtain the data set to be processed from the target database to retrain the target fault monitoring model. Through multiple iterations and structural optimization of the target fault monitoring model, the accuracy of the target fault monitoring model in fault monitoring of industrial equipment based on the operating data of the industrial equipment is improved, and the universality and logical robustness are enhanced.

[0078] In a specific embodiment, the device may further include:

[0079] The data file uploading module is used to store the industrial equipment fault data by using the pre-created target data set object, and upload the target data set object to the target database in the form of an offline data file.

[0080] In a specific embodiment, the data set construction module 11 may specifically include:

[0081] A data set processing unit is used to standardize the data set to be processed using a target data standardization operator to obtain standardized data, and to clean the standardized data using a target data cleaning operator to obtain the target data; the target data standardization operator includes a string to date operator and a column filtering operator, and the target data cleaning operator includes a timestamp alignment operator and a missing value processing operator.

[0082] In a specific embodiment, the data set construction module 11 may specifically include:

[0083] A data denoising unit, used to determine matching data corresponding to each column of data in the target data, and compare each column of data with the corresponding matching data using a column aggregation operator to remove noise data in the target data;

[0084] The data partitioning unit is used to partition the target data after noise data is removed using a data partitioning operator to construct the target training data set and the target test data set.

[0085] In a specific embodiment, the device may further include:

[0086] A model publishing module is used to upload the target fault monitoring model to the target model management library if the output result meets the model publishing conditions, so as to download the target fault monitoring model from the target model management library through the model edge reasoning end and deploy the target fault monitoring model locally on the model edge reasoning end.

[0087] In a specific embodiment, the device may further include:

[0088] A model update module is used to obtain feedback information uploaded to the target database by the model edge inference end, optimize the target fault monitoring model based on the feedback information to obtain a new target fault monitoring model, and jump to the step of uploading the target fault monitoring model to the target model management library.

[0089] Furthermore, the present application also discloses an electronic device. Fig.12 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content in the diagram cannot be regarded as any limitation on the scope of use of the present application.

[0090] Fig.12 A schematic diagram of the structure of an electronic device 20 provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the method for constructing an industrial equipment fault monitoring model disclosed in any of the aforementioned embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0091] In this embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0092] In addition, the memory 22 as a carrier for resource storage may be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon may include an operating system 221, a computer program 222, etc., and the storage method may be temporary storage or permanent storage.

[0093] The operating system 221 is used to manage and control the hardware devices on the electronic device 20 and the computer program 222, which can be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program that can be used to complete the industrial equipment fault monitoring model construction method performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program that can be used to complete other specific tasks.

[0094] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the aforementioned disclosed industrial equipment fault monitoring model construction method is implemented. The specific steps of the method can refer to the corresponding contents disclosed in the aforementioned embodiments, and will not be repeated here.

[0095] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0096] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0097] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0098] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0099] The technical solution provided by the present application is introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technicians in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for constructing an industrial equipment fault monitoring model, characterized in that: include: Using a target data set reading operator to obtain a data set to be processed from a target database, using a target data processing operator to process the data set to be processed to obtain target data, and using a target data set construction operator to construct a target training data set and a target test data set based on the target data; The target database is a database for recording industrial equipment fault data; Inputting the target training data set into a target fault monitoring model to train the target fault monitoring model, inputting the target test data set into a new target fault monitoring model obtained through training, and determining whether an output result of the target fault monitoring model meets a model release condition; If the output result does not meet the model publishing condition, determine whether the output result meets the preset model structure optimization condition, and when the output result does not meet the preset model structure optimization condition, perform a model iteration operation to jump to the step of obtaining the to-be-processed data set from the target database using the target data set reading operator; If the output result satisfies the preset model structure optimization condition, a model structure optimization operation is performed to optimize the model structure of the target fault monitoring model, and the process jumps to the step of obtaining the data set to be processed from the target database using the target data set reading operator; Among them, the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visualization modeling platform, and the target fault monitoring model is a model constructed through the target visualization modeling platform.

2. The method for constructing an industrial equipment fault monitoring model according to claim 1, characterized in that: Before using the target data set reading operator to obtain the data set to be processed from the target database, the method further includes: The industrial equipment fault data is stored using a pre-created target data set object, and the target data set object is uploaded to the target database in the form of an offline data file.

3. The method for constructing an industrial equipment fault monitoring model according to claim 1, characterized in that: The using the target data processing operator to process the data set to be processed to obtain target data includes: The target data standardization operator is used to standardize the data set to be processed to obtain standardized data, and the target data cleaning operator is used to clean the standardized data to obtain the target data; the target data standardization operator includes a string to date operator and a column filtering operator, and the target data cleaning operator includes a timestamp alignment operator and a missing value processing operator.

4. The method for constructing an industrial equipment fault monitoring model according to claim 1, characterized in that: The method of constructing a target training dataset and a target test dataset based on the target data by using a target dataset construction operator includes: Determine matching data corresponding to each column of data in the target data, and compare each column of data with the corresponding matching data using a column aggregation operator to remove noise data in the target data; The target data after noise data is removed is partitioned using a data partitioning operator to construct the target training data set and the target test data set.

5. The method for constructing an industrial equipment fault monitoring model according to claim 1, characterized in that: The model publishing condition is that the accuracy of the output result is greater than or equal to a first accuracy threshold, and the preset model structure optimization condition is that the accuracy of the output result is less than a second accuracy threshold, and the second accuracy threshold is less than the first accuracy threshold.

6. The method for constructing an industrial equipment fault monitoring model according to any one of claims 1 to 5, characterized in that: Also includes: If the output result meets the model publishing conditions, the target fault monitoring model is uploaded to the target model management library, so that the target fault monitoring model can be downloaded from the target model management library through the model edge reasoning end and deployed locally on the model edge reasoning end.

7. The method for constructing an industrial equipment fault monitoring model according to claim 6, characterized in that: Also includes: Obtain feedback information uploaded to the target database by the model edge inference end, optimize the target fault monitoring model based on the feedback information to obtain a new target fault monitoring model, and jump to the step of uploading the target fault monitoring model to the target model management library.

8. An industrial equipment fault monitoring model construction device, characterized in that: include: A data set construction module, used to obtain a data set to be processed from a target database using a target data set reading operator, process the data set to be processed using a target data processing operator to obtain target data, and construct a target training data set and a target test data set based on the target data using a target data set construction operator; The target database is a database for recording industrial equipment fault data; A model training module is used to input the target training data set into a target fault monitoring model to train the target fault monitoring model, input the target test data set into a new target fault monitoring model obtained through training, and determine whether an output result of the target fault monitoring model meets a model release condition; A model iteration module, configured to determine whether the output result satisfies a preset model structure optimization condition if the output result does not satisfy the model publishing condition, and to perform a model iteration operation to jump to the step of obtaining a to-be-processed data set from a target database using a target data set reading operator when the output result does not satisfy the preset model structure optimization condition; A model optimization module, configured to perform a model structure optimization operation to optimize the model structure of the target fault monitoring model if the output result satisfies the preset model structure optimization condition, and jump to the step of obtaining a to-be-processed data set from a target database using a target data set reading operator; Among them, the target data set reading operator, the target data processing operator and the target data set construction operator are operators selected through the target visualization modeling platform, and the target fault monitoring model is a model constructed through the target visualization modeling platform.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the method for constructing an industrial equipment fault monitoring model according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the method for constructing an industrial equipment fault monitoring model according to any one of claims 1 to 7 is implemented.

Images