Time sequence analysis fault prediction method for user side terminal and control loop

Abstract

The invention discloses a time sequence analysis fault prediction method for a user side terminal and a control loop, and belongs to the technical field of power system fault processing, and the method comprises the steps: collecting the operation data, fault logs and maintenance records of the user side terminal and the control loop through a sensor and a data collection device; preprocessing the data; analyzing the data by using an Apriori algorithm, and mining an association rule between an equipment fault and a running state; performing time sequence analysis by using LSTM network model data; and finally carrying out maintenance strategy optimization. According to the time sequence analysis fault prediction method for the user side terminal and the control loop, the Apriori algorithm is creatively combined with the LSTM model, the equipment fault can be predicted and responded in real time, the maintenance plan and the fault response strategy can be adjusted according to the prediction result, and the fault prediction efficiency is improved. Therefore, the equipment operation efficiency is optimized, the fault downtime is reduced, and stable operation and efficient management of a power system are guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of power system fault handling, and particularly to a time series analysis fault prediction method for user-side terminals and control circuits. Background Art

[0002] With the rapid development of global information technology and artificial intelligence, data-driven decision-making and algorithms are playing an increasingly important role in power system management. New power load management systems especially need to be able to process and analyze large amounts of data to ensure the efficient and stable operation of the system. However, in the face of increasingly complex power networks and growing energy demands, especially in these key parts such as user-side terminals and control circuits, the system's data processing and fault prediction technologies are facing major challenges.

[0003] User-side terminals and control circuits are the core to ensure the performance of the power load management system, and their stability and reliability directly affect the operating efficiency of the entire power system. Real-time monitoring and fault prediction are particularly crucial in these links because this can significantly reduce the system's downtime and improve operating efficiency. However, traditional fault detection methods often rely on a large amount of historical data and empirical rules, and their applicability and effectiveness are limited in new systems or low-data environments.

[0004] As two powerful data analysis technologies, association rule mining and time series analysis have respectively demonstrated their unique advantages in the fields of pattern recognition and predictive analysis. Association rules can reveal the dependency relationships and fault propagation paths between devices, while time series analysis can predict the future state and potential fault points of devices. However, how to effectively combine these two technologies and apply them to the power load management system to achieve more accurate and real-time fault prediction and system maintenance remains a technical problem to be solved.

[0005] Therefore, developing a fault prediction and maintenance method that combines association rule mining and time series analysis to optimize the performance of user-side terminals and control circuits in the power load management system is of great significance for ensuring the stability and efficient operation of the system. Summary of the Invention

[0006] The object of the present invention is to provide a time series analysis fault prediction method for user-side terminals and control circuits, which can predict and respond to equipment faults in real time, adjust the maintenance plan and fault response strategy according to the prediction results, improve the equipment operation efficiency and reduce the fault downtime, and ensure the stable operation and efficient management of the power system.

[0007] To achieve the above object, the present invention provides a time series analysis fault prediction method for user-side terminals and control circuits, including the following steps:

[0008] S1. Collect the operation data, fault logs, and maintenance records of the user-side terminal and control loop;

[0009] S2. Preprocess the collected data;

[0010] S3. Mine association rules from the preprocessed data to determine the target association rules;

[0011] S4. Train the LSTM network model using the preprocessed data, and use the trained LSTM network model to predict the real-time data to obtain the target faults that will occur in the future;

[0012] S5. Optimize the maintenance strategy based on the target association rules and the target faults that will occur in the future.

[0013] Preferably, in S2, the preprocessing includes the following steps:

[0014] S21. Data cleaning;

[0015] Remove outliers and missing data;

[0016] S22. Normalize numerical data;

[0017] Normalize the numerical data in the data processed in S21. The calculation formula for normalization is:

[0018]

[0019] S23. Vectorize non-numerical data;

[0020] For the non-numerical data in the data processed in S21, use the word embedding technology to perform vectorization processing;

[0021] S24. Data merging;

[0022] Merge the normalized numerical data processed in S22 and the vectorized non-numerical data processed in S23 into a unified data set.

[0023] Preferably, in S3, the association rule mining is specifically: use the Apriori algorithm to analyze the data preprocessed in S2 to mine the association rules between equipment faults and operating states, and screen out the association rules valuable for fault prediction as the target association rules by calculating the support and confidence of each association rule.

[0024] Preferably, in S3, the association rule mining includes the following steps:

[0025] S31. Construct frequent item sets;

[0026] S311. Preliminary screening: Extract all combinations of device states and parameters from the preprocessed data and calculate the support degree.

[0027] S312. Threshold-based screening: Eliminate combinations of device states and parameters with a support degree lower than the set threshold, and only retain combinations with a high support degree.

[0028] S313. Construct frequent item sets: Generalize combinations of device states and parameters with a support degree higher than the set threshold into frequent item sets for subsequent association rule mining.

[0029] S32. Extract association rules.

[0030] Derive rules with a confidence degree higher than the confidence threshold from the frequent item sets determined in S3 as the target association rules.

[0031] Preferably, in S4, the specific process of training the LSTM network model using the preprocessed data is as follows: Apply the LSTM network model to analyze the preprocessed data in S2 to learn the time dependence and dynamic change patterns in the preprocessed data.

[0032] Preferably, in S5, the optimization and maintenance strategy includes the following steps:

[0033] S51. Adjust the maintenance plan;

[0034] Combined with the target association rules and the target faults that will occur in the future, rearrange the maintenance cycle and maintenance content of the equipment;

[0035] S52. Develop a fault response strategy;

[0036] The target association rules include device fault modes. Based on the device fault modes obtained in S3 and the target faults that will occur in the future obtained in S4, develop a fault response strategy. The fault response strategy includes formulating emergency measures, spare parts, and technical support resources when a fault occurs.

[0037] Preferably, a time series analysis fault prediction system for a user-side terminal and a control loop is characterized by including:

[0038] A data acquisition module for collecting and storing the operation data, fault logs, and maintenance records of the user-side terminal and the control loop;

[0039] A data preprocessing module for preprocessing and saving the data in the data acquisition module;

[0040] An association rule mining module for performing association rule mining on the data preprocessed by the data preprocessing module;

[0041] A time series analysis module for learning the time dependence in the data preprocessed by the data preprocessing module;

[0042] A maintenance strategy optimization module for optimizing the maintenance strategy based on the results of the association rule mining module and the time series analysis module.

[0043] Preferably, the association rule mining module and the time series analysis module process the output content of the data preprocessing module synchronously.

[0044] Preferably, a computer device, characterized in that it includes at least one processor; the processor is used to store one or more programs; when the one or more programs are executed by the one or more processors, the time series analysis fault prediction method of the user-side terminal and the control loop described above is implemented.

[0045] Preferably, a computer-readable storage medium, characterized in that a computer program is stored thereon, and when the computer program is executed, the time series analysis fault prediction method of the user-side terminal and the control loop described above is implemented.

[0046] Therefore, the present invention adopts the above-mentioned time series analysis fault prediction method for a user-side terminal and a control loop, combines the Apriori algorithm with the LSTM model, uses the Apriori algorithm to analyze the maintenance records and fault logs of the device, mines the association rules between faults and specific operating states, and improves the recognition accuracy of fault patterns; learns the time dependence in the historical data of the device through the LSTM model, predicts the future changes in the device state, responds to potential faults in advance, thereby optimizing the device operation efficiency and reducing the fault downtime, and ensuring the stable operation and efficient management of the power system.

[0047] The technical solution of the present invention will be further described in detail below with reference to the drawings and embodiments. Description of the Drawings

[0048] Figure 1 It is a flowchart of the method of an embodiment of the time series analysis fault prediction method for a user-side terminal and a control loop of the present invention. Detailed Embodiments

[0049] The technical solution of the present invention will be further described below with reference to the drawings and embodiments.

[0050] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings as understood by those of ordinary skill in the art to which the present invention pertains. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. Words such as "comprising" or "including" mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Upper", "lower", "left", "right", etc. are only used to indicate relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0051] Embodiment 1

[0052] The present invention provides a time series analysis fault prediction method for a user-side terminal and a control loop, and the flow of the prediction method is as Figure 1 shown, including the following steps:

[0053] S1. Data collection;

[0054] Use sensors and data collection devices to collect the operation data, fault logs and maintenance records of the user-side terminal and the control loop.

[0055] S2. Preprocess the collected data;

[0056] In order to improve the quality and efficiency of analysis, preprocess the collected data, including the following steps:

[0057] S21. Data cleaning;

[0058] Remove outliers and missing data to ensure the integrity and accuracy of the data.

[0059] S22. Normalize numerical data;

[0060] Normalize the numerical data in the data processed in S21 for further processing by the model. The calculation formula for normalization is:

[0061]

[0062] S23. Vectorize non-numerical data;

[0063] For the non-numerical data in the data processed in S21, such as the text in maintenance records and fault logs, use word embedding techniques such as Word2Vec for vectorization processing to convert it into a format suitable for machine learning models.

[0064] S24. Data merging;

[0065] Merge the numerical data after normalization in S22 and the non - numerical data after vectorization in S23 into a unified dataset for association rule mining.

[0066] S3. Perform association rule mining on the pre - processed data to determine the target association rules;

[0067] Use the Apriori algorithm to analyze the pre - processed data in S2 to mine the association rules between equipment failures and operating states. For example, identify the relationships between high - temperature and high - current states and specific failures. By calculating the support and confidence of each association rule, screen out the rules valuable for fault prediction and identify potential equipment failure modes, which specifically include the following steps:

[0068] S31. Construct frequent item sets;

[0069] Set a support threshold. Based on the set support threshold, analyze and identify the combinations of equipment states and parameters that frequently occur simultaneously in the pre - processed data in S2. When the proportion of the number of times the combination of equipment states and parameters appears in the entire dataset exceeds the set support threshold, it is regarded as a frequent item. Use the frequently co - occurring combinations of equipment states and parameters to construct frequent item sets, which specifically include the following steps:

[0070] S311. Preliminary screening: Extract all combinations of equipment states and parameters from the pre - processed data and calculate the support;

[0071] S312. Threshold - setting screening: Eliminate the combinations of equipment states and parameters with support lower than the set threshold, and only retain the combinations with high support;

[0072] S313. Construct frequent item sets: Generalize the combinations of equipment states and parameters with support higher than the set threshold into frequent item sets for subsequent association rule mining.

[0073] S32. Extract association rules;

[0074] Set a confidence threshold. When the confidence of a frequent item association exceeds the set confidence threshold, it belongs to high confidence. Derive the rules with confidence higher than the confidence threshold from the frequent item sets determined in S31 as the target association rules to reveal the relationships between equipment failures and operating parameters.

[0075] S4. Time - series analysis;

[0076] Apply the LSTM network model to analyze the pre - processed data in S2, learn the time - dependence in the pre - processed historical data, and predict the trends of key parameters and potential fault points, specifically as follows:

[0077] S41. Model construction;

[0078] Use the preprocessed data in S2 to train the LSTM network model, and the LSTM network model learns and understands the time-dependent relationships and dynamic change patterns in the preprocessed data.

[0079] S42. Fault prediction;

[0080] Use the trained LSTM network model to analyze the real-time data and predict possible future faults or abnormal states.

[0081] S5. Optimize the maintenance strategy based on the target association rules and the target faults that will occur in the future;

[0082] Combine the target association rules and the target faults that will occur in the future to achieve real-time fault prediction and response, optimize the equipment operation efficiency and reduce the fault downtime. Specifically:

[0083] S51. Adjust the maintenance plan;

[0084] Combine the target association rules and the target faults that will occur in the future to obtain the possible future fault time points of the equipment, and re-arrange the maintenance cycle and maintenance content of the equipment. For example, if the prediction result shows that a certain control loop may fail in the near future, then the inspection and necessary maintenance of this loop should be planned in advance to prevent the occurrence of faults.

[0085] S52. Develop a fault response strategy;

[0086] The target association rules include the equipment fault modes. Based on the equipment fault modes obtained from S3 and the target faults that will occur in the future obtained from S4, develop a fault response strategy. The fault response strategy includes formulating emergency measures, spare parts, and technical support resources when a fault occurs. The fault response strategy aims to minimize the system downtime caused by faults and quickly restore the normal operation state of the system.

[0087] The present invention also provides a time series analysis fault prediction system for a user-side terminal and a control loop. The modules and functions are as follows:

[0088] A data acquisition module for collecting and storing the operation data, fault logs, and maintenance records of the user-side terminal and the control loop;

[0089] A data preprocessing module for preprocessing and saving the data in the data acquisition module;

[0090] An association rule mining module for mining association rules from the data preprocessed by the data preprocessing module;

[0091] A time series analysis module for learning the time dependence in the data preprocessed by the data preprocessing module;

[0092] A maintenance strategy optimization module for optimizing the maintenance strategy based on the results of the association rule mining module and the time series analysis module.

[0093] In this system, the association rule mining module and the time series analysis module process the output content of the data preprocessing module synchronously.

[0094] The present invention also relates to a computer device and a readable storage medium. The computer device includes at least one processor; the processor is used for storing one or more programs; when the one or more programs are executed by the one or more processors, a time series analysis fault prediction method for a user-side terminal and a control loop according to the present invention is implemented. The storage medium stores a computer program, and when the computer program is executed, a time series analysis fault prediction method for a user-side terminal and a control loop according to the present invention is implemented.

[0095] Therefore, by adopting the above time series analysis fault prediction method for a user-side terminal and a control loop, the present invention creatively combines the Apriori algorithm with the LSTM model to improve the fault prediction efficiency of the user-side terminal and the control loop of the power system. By using the Apriori algorithm to analyze the maintenance records and fault logs of the equipment, the association rules between faults and specific operating states are mined, and the recognition accuracy of fault patterns is improved. At the same time, by using the LSTM model to learn the time dependence in the historical data of the equipment, the future changes of the equipment state are predicted to respond to potential faults in advance. This method can not only predict and respond to equipment faults in real time, but also adjust the maintenance plan and fault response strategy according to the prediction results, thereby optimizing the equipment operation efficiency and reducing the fault downtime, and supporting the stable operation and efficient management of the power system.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for predicting faults by time series analysis of user-side terminals and control loops, characterized in that: The following steps are involved: S1. Collect operation data, fault logs and maintenance records of user-side terminals and control loops; S2, preprocessing the collected data; S3, performing association rule mining on the preprocessed data to determine the target association rules; S4. Use the preprocessed data to train the LSTM network model, and use the trained LSTM network model to predict the real-time data to obtain the target faults that will occur in the future; S5. Optimize maintenance strategies based on target association rules and future target failures.

2. According to claim 1, a method for predicting faults by time series analysis of a user-side terminal and a control loop is characterized in that: In S2, the preprocessing includes the following steps: S21, data cleaning; Remove outliers and missing data; S22, normalization of numerical data; The numerical data in the data processed by S21 is normalized, and the calculation formula for normalization is: S23, vectorization of non-numerical data; The non-numerical data in the data processed by S21 is vectorized using word embedding technology; S24, data merging; The numerical data normalized by S22 and the non-numerical data vectorized by S23 are merged into a unified data set.

3. The method for predicting faults by time series analysis of a user-side terminal and a control loop according to claim 1, characterized in that: In S3, association rule mining is specifically as follows: using the Apriori algorithm to analyze the preprocessed data in S2 to mine the association rules between equipment failures and operating status, and by calculating the support and confidence of each association rule, selecting the association rules that are valuable for fault prediction as the target association rules.

4. The method for predicting faults by time series analysis of a user-side terminal and a control loop according to claim 1, characterized in that: In S3, association rule mining includes the following steps: S31, construct frequent itemsets; S311, preliminary screening, extracting all device states and parameter combinations from the preprocessed data and calculating the support; S312, setting a threshold for screening, eliminating device state and parameter combinations with support values ​​lower than the set threshold, and only retaining combinations with high support values; S313, constructing frequent item sets, summarizing the device status and parameter combinations with support higher than a set threshold as frequent item sets for subsequent association rule mining; S32, extracting association rules; From the frequent item sets determined in S31, rules with confidence levels higher than a confidence threshold are derived as target association rules.

5. The method for predicting faults by time series analysis of a user-side terminal and a control loop according to claim 1, characterized in that: In S4, the LSTM network model is trained using the preprocessed data. Specifically, the LSTM network model is applied to analyze the preprocessed data in S2 to learn the time dependency and dynamic change patterns in the preprocessed data.

6. The method for predicting faults by time series analysis of a user-side terminal and a control loop according to claim 1, characterized in that: In S5, the optimization maintenance strategy includes the following steps: S51. Adjust the maintenance plan; Rearrange the equipment maintenance cycle and maintenance content based on target association rules and future target faults; S52. Formulate a fault response strategy; The target association rules include equipment failure modes. Based on the equipment failure modes obtained in S3 and the target failures that will occur in the future obtained in S4, a failure response strategy is formulated. The failure response strategy includes formulating emergency measures, spare parts, and technical support resources when a failure occurs.

7. A time series analysis fault prediction system for user-side terminals and control loops, characterized in that: include: Data acquisition module, used to collect and store operation data, fault logs and maintenance records of user-side terminals and control loops; A data preprocessing module, used for preprocessing and saving the data in the data acquisition module; An association rule mining module, used for performing association rule mining on the data preprocessed by the data preprocessing module; A time series analysis module, used for learning the time dependency in the data preprocessed by the data preprocessing module; The maintenance strategy optimization module optimizes the maintenance strategy according to the results of the association rule mining module and the time series analysis module.

8. A time series analysis fault prediction system for user-side terminals and control loops according to claim 7, characterized in that: The association rule mining module and the time series analysis module synchronously process the output content of the data preprocessing module.

9. A computer device, characterized in that: It includes at least one processor; the processor is used to store one or more programs; when the one or more programs are executed by the one or more processors, a time series analysis fault prediction method for a user-side terminal and a control loop as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, a time series analysis fault prediction method for a user-side terminal and a control loop as described in any one of claims 1 to 8 is implemented.

Images