A DC Engineering Intelligent Maintenance System

By combining data acquisition, intelligent learning, fault diagnosis and prediction modules with sensor and deep learning technologies, automated maintenance of DC engineering has been achieved, solving the problems of low efficiency and poor accuracy of traditional manual maintenance, and providing a fast and accurate fault handling solution.

CN123073766APending Publication Date: 2026-11-10GUANGZHOU BUREAU CSG EHV POWER TRANSMISSION
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611123845.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-28
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Traditional DC engineering maintenance relies on manual inspections, which is inefficient, its accuracy is greatly affected by subjective factors, it cannot predict and handle faults in a timely manner, and it is difficult to quickly detect potential equipment problems.

Method used

The system employs a data acquisition module to acquire equipment and line data, an intelligent learning module to analyze fault knowledge, a fault diagnosis module to diagnose anomalies, a fault prediction module to predict faults, a fault repair module to provide repair strategies, and a data storage module to store data. It utilizes technologies such as sensors, convolutional neural networks, and long short-term memory networks to achieve automated repair.

Benefits of technology

It enables automated fault diagnosis and prediction for DC equipment and lines, provides rapid maintenance strategies, reduces labor costs and equipment damage risks, and improves maintenance efficiency and accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123073766A_ABST
    Figure CN123073766A_ABST
Patent Text Reader

Abstract

This invention provides a smart maintenance system for DC engineering projects. The system includes: a data acquisition module for collecting data from various DC equipment and DC transmission lines; a smart learning module for understanding and analyzing fault types and fault repair knowledge; a fault diagnosis module for analyzing abnormal feature data in the data of various DC equipment and DC transmission lines, and diagnosing the fault types of each DC equipment and DC transmission line based on the abnormal feature data; a fault prediction module for constructing a fault prediction model and predicting faults in each DC equipment and DC transmission line through the fault prediction model; a fault repair module for providing fault repair strategies for maintenance personnel; and a data storage module for storing data from various DC equipment and DC transmission lines. This invention can quickly detect potential faults in DC engineering projects, avoid sudden failures and serious damage, and provide maintenance personnel with repair strategies, reducing equipment maintenance and replacement costs.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of DC engineering maintenance technology, and in particular to a smart DC engineering maintenance system. Background Technology

[0002] Traditional DC engineering maintenance mainly relies on manual inspections and experience-based judgment. Maintenance personnel need to go to the site regularly and judge whether there are any abnormalities in DC equipment and transmission lines based on their own opinions or using simple tools. The maintenance efficiency is extremely low. Faced with a large and complex DC project that covers a large number of equipment and lines, maintenance personnel need to spend a lot of time and energy to check them one by one. Moreover, it is impossible to predict the faults of DC equipment and transmission lines in advance, and it is not possible to remind maintenance personnel to carry out maintenance in a timely manner. Furthermore, if a fault occurs during the inspection interval, it is difficult to detect and deal with it in a timely manner. At the same time, the accuracy of manual judgment is greatly affected by subjective factors. The technical level, work experience, and mental state of different maintenance personnel can all lead to differences in judgment results. Summary of the Invention

[0003] Therefore, the purpose of this invention is to provide a smart maintenance system for DC engineering to solve or at least partially solve the above-mentioned problems existing in the prior art.

[0004] To achieve the above objectives, the present invention provides a smart maintenance system for DC engineering projects, the system comprising: Data acquisition module: used to collect data from various DC devices and DC transmission lines. The DC device data includes DC voltage, DC current, power, load rate, and operating temperature. The DC transmission line data includes line DC voltage, line DC current, and line temperature. The intelligent learning module is used to acquire fault types and fault repair knowledge of various DC equipment and DC transmission lines, understand and analyze the acquired fault types and fault repair knowledge, and integrate them to generate a fault repair dataset. Fault diagnosis module: used to analyze abnormal feature data in the data of various DC equipment and DC transmission lines, and diagnose the fault type of each DC equipment and DC transmission line based on the abnormal feature data; Fault prediction module: Used to acquire historical fault data and real-time operating data of each DC equipment and DC transmission line, and to build a fault prediction model based on the acquired data, and to predict faults of each DC equipment and DC transmission line through the fault prediction model; Fault Repair Module: Based on the fault diagnosis and fault prediction results, combined with the intelligent learning module, it provides fault repair strategies for maintenance personnel. Data storage module: Used to store real-time operating data, historical fault data and maintenance records of various DC equipment and DC transmission lines.

[0005] Furthermore, the data acquisition module is equipped with various types of sensors at fixed locations on each DC device and DC transmission line. These sensors are used to collect various operating parameters of each DC device and DC transmission line and transmit these operating parameters to other modules via wired or wireless communication.

[0006] Furthermore, the intelligent learning module specifically includes the following steps: S11. Collect existing fault types and fault repair knowledge of various DC equipment and DC transmission lines, including documents in various formats. S12. Convert documents of multiple formats to the same format and perform preprocessing operations on them. Split the preprocessed documents into long documents according to semantic boundaries or fixed lengths, while retaining the information of each segment. S13. Input the split long document and the unsplit document into the large language model, and use the large language model to extract structured data from them. S14. An embedded model is introduced into the large language model. The embedded model converts structured and unstructured data into high-dimensional vectors. The embedded model learns the correlation between fault types and fault repair knowledge through high-dimensional vectors, and maps similar data to similar embedding spaces, so that the large language model can effectively understand and analyze based on the embedding space.

[0007] Furthermore, the fault diagnosis module specifically includes the following steps: S21. Based on the data of each DC device and DC transmission line, preset the corresponding data anomaly thresholds respectively; S22. Obtain dynamic data of each DC device and DC transmission line from the historical time to the current time, and analyze the abnormal feature data of each DC device and DC transmission line in the dynamic data according to the preset data abnormality threshold. S23. Perform preprocessing operations on the abnormal feature data and construct a convolutional neural network model; S24. Initialize the parameters of the global optimization algorithm based on the abnormal feature data after preprocessing, and update the fitness function value and optimal feasible solution of the global optimization algorithm based on the initialized parameters. S25. Determine whether the fitness function value and the optimal feasible solution of the updated global optimization algorithm in step S24 satisfy the stopping condition of the iteration. If not, repeat step S24. If they are satisfied, assign the optimal parameter combination obtained by the global optimization algorithm to the convolutional neural network model constructed in step S23. S26. Using the preprocessed abnormal feature data and the fault repair dataset, train the convolutional neural network model after the assignment in step S25 to generate a fault diagnosis model. Use the fault diagnosis model to diagnose the fault types of each DC equipment and DC transmission line.

[0008] Furthermore, the fault prediction module specifically includes the following steps: S31. Obtain historical fault dynamic data of each DC equipment and DC transmission line, and perform preprocessing operations on it to obtain historical fault dynamic data after dimensionality reduction. S32. Set a sliding window of a certain length and slide it on the historical fault dynamic data after dimensionality reduction at a certain unit time interval, and divide the historical fault dynamic data into training set and test set according to the time point. S33. Construct a fault prediction model, perform regularization on the training set, and train the fault prediction model using the regularized training set. S34. Introduce a bidirectional long short-term memory network into the fault prediction model. The long short-term memory network is used to capture the correlation information of different time points in the historical fault dynamic data, and randomly set the output of some training data in the training set to 0 to prevent the fault prediction model from overfitting. S35. Convert the training data in the training set processed by the bidirectional long short-term memory network in step S34 into one-dimensional data, and retain the features of the training data during the conversion process. Then, convert the converted training data into a fully connected layer. S36. Use a fully connected layer to integrate the features of the training data, and combine the local features of the training data into a global representation through weighted combination. S37. Input the training data processed by the fully connected layer into the LSTM layer. Set the parameter optimizer through the LSTM layer during the training process of the fault prediction model. Optimize the loss function of the fault prediction model through the parameter optimizer to complete the training of the fault prediction model. S38. Input the test set into the trained fault prediction model to verify the fault detection performance of the model, and input the real-time running data into the model to perform fault prediction for each DC equipment and DC transmission line.

[0009] Furthermore, the fault diagnosis module includes: The DC transmission line adopts a bipolar loop connection method for DC transmission. If a fault occurs or there is a risk of fault in one pole of the DC transmission line, the DC transmission line is converted to a single-pole grounding loop operation mode for DC transmission with the goal of optimizing the cost of fault repair and operation loss of the DC transmission line. At the same time, the pole with fault or risk of fault is repaired by combining the fault repair data set. If a DC device malfunctions or is at risk of malfunction, the fault type of the DC device is obtained, and the fault type is matched with the fault repair dataset to determine its correlation. The repair knowledge with the highest correlation after understanding and analysis is selected, and a fault repair strategy is generated for the repair personnel based on the repair knowledge.

[0010] Furthermore, the data storage module adopts a hierarchical storage architecture, which consists of a real-time database layer and a historical database layer. The real-time database layer is used to store real-time operating parameters, and the historical database layer is used to store historical fault data and maintenance records.

[0011] Compared with the prior art, the beneficial effects of the present invention are: This invention proposes a smart maintenance system for DC engineering projects. It collects data from various DC equipment and transmission lines through a data acquisition module to ensure data authenticity and accuracy. A smart learning module enables in-depth understanding and analysis of DC maintenance engineering knowledge. A fault diagnosis module diagnoses faults in various DC equipment and transmission lines. A fault prediction module predicts faults in advance. A fault repair module provides maintenance strategies based on fault type. A data storage module stores various types of data from DC equipment and transmission lines. This invention can quickly identify potential faults in DC engineering projects, avoid sudden failures and serious damage, and provide maintenance strategies to reduce equipment repair and replacement costs. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only preferred embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of a DC engineering intelligent maintenance system provided in an embodiment of the present invention. Detailed Implementation

[0014] The principles and features of the present invention are described below with reference to the accompanying drawings. The listed embodiments are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0015] Reference Figure 1 This embodiment provides a smart maintenance system for DC engineering, the system comprising: Data acquisition module: Used to collect data from various DC devices and DC transmission lines. The DC device data includes DC voltage, DC current, power, load rate, and operating temperature. The DC transmission line data includes line DC voltage, line DC current, and line temperature. Specifically, it includes: The data acquisition module is equipped with various types of sensors at fixed locations on each DC device and DC transmission line. These sensors are used to collect various operating parameters of each DC device and DC transmission line and transmit these operating parameters to other modules via wired or wireless communication.

[0016] The intelligent learning module is used to acquire fault types and troubleshooting knowledge for various DC equipment and DC transmission lines. It then understands, analyzes, and integrates this acquired fault types and troubleshooting knowledge to generate a fault troubleshooting dataset. Specifically, it includes the following steps: S11. Collect existing fault types and fault repair knowledge of various DC equipment and DC transmission lines, including documents in various formats. S12. Convert documents of multiple formats to the same format and perform preprocessing operations on them. Split the preprocessed documents into long documents according to semantic boundaries or fixed lengths, while retaining the information of each segment. S13. Input the split long document and the unsplit document into the large language model, and use the large language model to extract structured data from them. S14. An embedded model is introduced into the large language model. The embedded model converts structured and unstructured data into high-dimensional vectors. The embedded model learns the correlation between fault types and fault repair knowledge through high-dimensional vectors, and maps similar data to similar embedding spaces, so that the large language model can effectively understand and analyze based on the embedding space.

[0017] Fault Diagnosis Module: Used to analyze abnormal feature data in the data of various DC equipment and DC transmission lines, and diagnose the fault type of each DC equipment and DC transmission line based on the abnormal feature data. Specifically, it includes the following steps: S21. Based on the data of each DC device and DC transmission line, preset the corresponding data anomaly thresholds respectively; S22. Obtain dynamic data of each DC device and DC transmission line from the historical time to the current time, and analyze the abnormal feature data of each DC device and DC transmission line in the dynamic data according to the preset data abnormality threshold. S23. Perform preprocessing operations on the abnormal feature data and construct a convolutional neural network model; S24. Initialize the parameters of the global optimization algorithm based on the abnormal feature data after preprocessing, and update the fitness function value and optimal feasible solution of the global optimization algorithm based on the initialized parameters. S25. Determine whether the fitness function value and the optimal feasible solution of the updated global optimization algorithm in step S24 satisfy the stopping condition of the iteration. If not, repeat step S24. If they are satisfied, assign the optimal parameter combination obtained by the global optimization algorithm to the convolutional neural network model constructed in step S23. S26. Using the preprocessed abnormal feature data and the fault repair dataset, train the convolutional neural network model after the assignment in step S25 to generate a fault diagnosis model. Use the fault diagnosis model to diagnose the fault types of each DC equipment and DC transmission line.

[0018] Fault prediction module: This module acquires historical fault data and real-time operational data for each DC device and DC transmission line, and constructs a fault prediction model based on the acquired data. The model then predicts faults for each DC device and DC transmission line. Specifically, it includes the following steps: S31. Obtain historical fault dynamic data of each DC equipment and DC transmission line, and perform preprocessing operations on it to obtain historical fault dynamic data after dimensionality reduction. S32. Set a sliding window of a certain length and slide it on the historical fault dynamic data after dimensionality reduction at a certain unit time interval, and divide the historical fault dynamic data into training set and test set according to the time point. S33. Construct a fault prediction model, perform regularization on the training set, and train the fault prediction model using the regularized training set. S34. Introduce a bidirectional long short-term memory network into the fault prediction model. The long short-term memory network is used to capture the correlation information of different time points in the historical fault dynamic data, and randomly set the output of some training data in the training set to 0 to prevent the fault prediction model from overfitting. S35. Convert the training data in the training set processed by the bidirectional long short-term memory network in step S34 into one-dimensional data, and retain the features of the training data during the conversion process. Then, convert the converted training data into a fully connected layer. S36. Use a fully connected layer to integrate the features of the training data, and combine the local features of the training data into a global representation through weighted combination. S37. Input the training data processed by the fully connected layer into the LSTM layer. Set the parameter optimizer through the LSTM layer during the training process of the fault prediction model. Optimize the loss function of the fault prediction model through the parameter optimizer to complete the training of the fault prediction model. S38. Input the test set into the trained fault prediction model to verify the fault detection performance of the model, and input the real-time running data into the model to perform fault prediction for each DC equipment and DC transmission line.

[0019] Fault Diagnosis Module: Based on fault diagnosis and prediction results, and in conjunction with the intelligent learning module, this module provides maintenance personnel with fault diagnosis strategies, specifically including: The DC transmission line uses a bipolar loop connection for DC power transmission. If a fault occurs or a fault risk exists in one pole of the DC transmission line, the DC transmission line is switched to a single-pole grounded loop operation mode to optimize the cost of fault repair and operation. Simultaneously, based on the fault repair data set, the pole with the fault or fault risk is repaired. The DC transmission status of the DC transmission line is represented as follows:

[0020] in, For DC transmission lines in The number of extremes during the time period This represents the total number of poles in a DC transmission line, which is typically bipolar and is taken as 2. This represents the number of poles currently operating on the DC transmission line. For DC transmission lines in Is the area under maintenance during this period? When =1, it indicates that the transmission line is in The period is under maintenance. When =0, it indicates that the transmission line is in The period is in normal operating condition; If a DC device malfunctions or is at risk of malfunction, the fault type of the DC device is obtained, and the fault type is matched with the fault repair dataset to determine its correlation. The repair knowledge with the highest correlation after understanding and analysis is selected, and a fault repair strategy is generated for the repair personnel based on the repair knowledge.

[0021] Data storage module: Used to store real-time operating data, historical fault data, and maintenance records of various DC equipment and DC transmission lines, specifically including: The data storage module adopts a hierarchical storage architecture, which consists of a real-time database layer and a historical database layer. The real-time database layer is used to store real-time operating parameters, and the historical database layer is used to store historical fault data and maintenance records.

[0022] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart maintenance system for DC engineering, characterized in that, The system includes: Data acquisition module: used to collect data from various DC devices and DC transmission lines. The DC device data includes DC voltage, DC current, power, load rate, and operating temperature. The DC transmission line data includes line DC voltage, line DC current, and line temperature. The intelligent learning module is used to acquire fault types and fault repair knowledge of various DC equipment and DC transmission lines, understand and analyze the acquired fault types and fault repair knowledge, and integrate them to generate a fault repair dataset. Fault diagnosis module: used to analyze abnormal feature data in the data of various DC equipment and DC transmission lines, and diagnose the fault type of each DC equipment and DC transmission line based on the abnormal feature data; Fault prediction module: Used to acquire historical fault data and real-time operating data of each DC equipment and DC transmission line, and to build a fault prediction model based on the acquired data, and to predict faults of each DC equipment and DC transmission line through the fault prediction model; Fault Repair Module: Based on the fault diagnosis and fault prediction results, combined with the intelligent learning module, it provides fault repair strategies for maintenance personnel. Data storage module: Used to store real-time operating data, historical fault data and maintenance records of various DC equipment and DC transmission lines.

2. The intelligent maintenance system for DC engineering according to claim 1, characterized in that, The data acquisition module is equipped with various types of sensors at fixed locations on each DC device and DC transmission line. These sensors are used to collect various operating parameters of each DC device and DC transmission line and transmit these operating parameters to other modules via wired or wireless communication.

3. The intelligent maintenance system for DC engineering according to claim 1, characterized in that, The intelligent learning module specifically includes the following steps: S11. Collect existing fault types and fault repair knowledge of various DC equipment and DC transmission lines, including documents in various formats. S12. Convert documents of multiple formats to the same format and perform preprocessing operations on them. Split the preprocessed documents into long documents according to semantic boundaries or fixed lengths, while retaining the information of each segment. S13. Input the split long document and the unsplit document into the large language model, and use the large language model to extract structured data from them. S14. An embedded model is introduced into the large language model. The embedded model converts structured and unstructured data into high-dimensional vectors. The embedded model learns the correlation between fault types and fault repair knowledge through high-dimensional vectors, and maps similar data to similar embedding spaces, so that the large language model can effectively understand and analyze based on the embedding space.

4. The intelligent maintenance system for DC engineering according to claim 1, characterized in that, The fault diagnosis module specifically includes the following steps: S21. Based on the data of each DC device and DC transmission line, preset the corresponding data anomaly thresholds respectively; S22. Obtain dynamic data of each DC device and DC transmission line from the historical time to the current time, and analyze the abnormal feature data of each DC device and DC transmission line in the dynamic data according to the preset data abnormality threshold. S23. Perform preprocessing operations on the abnormal feature data and construct a convolutional neural network model; S24. Initialize the parameters of the global optimization algorithm based on the abnormal feature data after preprocessing, and update the fitness function value and optimal feasible solution of the global optimization algorithm based on the initialized parameters. S25. Determine whether the fitness function value and the optimal feasible solution of the updated global optimization algorithm in step S24 satisfy the stopping condition of the iteration. If not, repeat step S24. If they are satisfied, assign the optimal parameter combination obtained by the global optimization algorithm to the convolutional neural network model constructed in step S23. S26. Using the preprocessed abnormal feature data and the fault repair dataset, train the convolutional neural network model after the assignment in step S25 to generate a fault diagnosis model. Use the fault diagnosis model to diagnose the fault types of each DC equipment and DC transmission line.

5. The intelligent maintenance system for DC engineering according to claim 1, characterized in that, The fault prediction module specifically includes the following steps: S31. Obtain historical fault dynamic data of each DC equipment and DC transmission line, and perform preprocessing operations on it to obtain historical fault dynamic data after dimensionality reduction. S32. Set a sliding window of a certain length and slide it on the historical fault dynamic data after dimensionality reduction at a certain unit time interval, and divide the historical fault dynamic data into training set and test set according to the time point. S33. Construct a fault prediction model, perform regularization on the training set, and train the fault prediction model using the regularized training set. S34. Introduce a bidirectional long short-term memory network into the fault prediction model. The long short-term memory network is used to capture the correlation information of different time points in the historical fault dynamic data, and randomly set the output of some training data in the training set to 0 to prevent the fault prediction model from overfitting. S35. Convert the training data in the training set processed by the bidirectional long short-term memory network in step S34 into one-dimensional data, and retain the features of the training data during the conversion process. Then, convert the converted training data into a fully connected layer. S36. Use a fully connected layer to integrate the features of the training data, and combine the local features of the training data into a global representation through weighted combination. S37. Input the training data processed by the fully connected layer into the LSTM layer. Set the parameter optimizer through the LSTM layer during the training process of the fault prediction model. Optimize the loss function of the fault prediction model through the parameter optimizer to complete the training of the fault prediction model. S38. Input the test set into the trained fault prediction model to verify the fault detection performance of the model, and input the real-time running data into the model to perform fault prediction for each DC equipment and DC transmission line.

6. The intelligent maintenance system for DC engineering according to claim 1, characterized in that, The fault diagnosis module includes: The DC transmission line adopts a bipolar loop connection method for DC transmission. If a fault occurs or there is a risk of fault in one pole of the DC transmission line, the DC transmission line is converted to a single-pole grounding loop operation mode for DC transmission with the goal of optimizing the cost of fault repair and operation loss of the DC transmission line. At the same time, the pole with fault or risk of fault is repaired by combining the fault repair data set. If a DC device malfunctions or is at risk of malfunction, the fault type of the DC device is obtained, and the fault type is matched with the fault repair dataset to determine its correlation. The repair knowledge with the highest correlation after understanding and analysis is selected, and a fault repair strategy is generated for the repair personnel based on the repair knowledge.

7. The intelligent maintenance system for DC engineering according to claim 1, characterized in that, The data storage module adopts a hierarchical storage architecture, which consists of a real-time database layer and a historical database layer. The real-time database layer is used to store real-time operating parameters, and the historical database layer is used to store historical fault data and maintenance records.