Substation production operation accident trip diagnosis method and system based on digital scene

Abstract

The invention discloses a transformer substation production operation accident trip diagnosis method and system based on a digital scene, belongs to the technical field of power system automation, and aims to solve the problems of low data processing efficiency, insufficient fault positioning accuracy, report generation lagging and insufficient intelligent level. Comprising the steps of multi-source data acquisition and preprocessing, multi-source data fusion model construction, accident trip diagnosis analysis model construction and application, accident trip report automatic generation, diagnosis result verification and model optimization. Full-process intelligent processing of transformer substation accident tripping is achieved through a closed-loop working mode and multi-module cooperative operation, when accident tripping happens to a transformer substation, the system automatically triggers a diagnosis process, key information is extracted from multi-source heterogeneous data, fault reasons are rapidly positioned through an intelligent algorithm, and a standardized diagnosis report is automatically generated. And the diagnosis model is iteratively optimized based on an actual maintenance result to form a continuously evolved intelligent diagnosis system.

Description

Technical Field

[0001] This invention belongs to the field of power system automation technology, specifically relating to a method and system for diagnosing tripping accidents in substations based on digital scenarios. Background Technology

[0002] In power system operation, substation fault tripping is a critical issue affecting power supply reliability. Traditional fault tripping diagnosis mainly relies on manual analysis, which has the following technical bottlenecks:

[0003] Low data processing efficiency: Substation operation data comes from diverse sources, including equipment telemetry and telemetry, meteorological information, defect records, and other heterogeneous data. Manual integration and analysis is time-consuming and labor-intensive, making it difficult to meet the needs of real-time diagnosis.

[0004] Insufficient accuracy in fault location: Traditional methods rely on the experience of maintenance personnel, which can easily lead to misjudgments in the analysis of complex faults, especially in multi-factor coupled fault scenarios, making it difficult to quickly and accurately identify the root cause of the trip.

[0005] Delayed report generation: After a power outage, the time required for manually writing diagnostic reports is long, which cannot provide timely decision support for fault handling and system recovery, and may lead to an expansion of the impact of the accident.

[0006] Insufficient intelligence level: It lacks in-depth mining of historical and real-time data, cannot achieve automatic learning and prediction of fault modes, and is difficult to adapt to the development needs of intelligent operation and maintenance of substations.

[0007] With the advancement of digital transformation of substations, real-time collection and intelligent analysis of massive operational data have become possible. However, how to efficiently integrate multi-source data and build intelligent diagnostic models remains a technical challenge that needs to be addressed.

[0008] Therefore, there is a need for a digital-based substation operation accident tripping diagnosis method and system to address the problems of low data processing efficiency, insufficient fault location accuracy, delayed report generation, and insufficient intelligence level in existing technologies. Summary of the Invention

[0009] The purpose of this invention is to provide a method and system for diagnosing tripping accidents in substations based on digital scenarios, so as to solve the problems mentioned in the background art.

[0010] To achieve the above objectives, the present invention provides the following technical solution: a method for diagnosing tripping accidents in substations based on digital scenarios, comprising the following steps:

[0011] S1. Multi-source data acquisition and preprocessing: Real-time acquisition of equipment telemetry and teleindication parameters, meteorological information, defect records, alarm notifications, on-site video monitoring data, information protection substation data and fault recording files before and after substation accident tripping; cleaning, classifying and summarizing the acquired data, integrating the data and mapping the associations to build a standardized accident tripping data pool.

[0012] S2. Construction of multi-source data fusion model: Based on the preprocessed data, feature extraction and dimensionality reduction algorithms are used to extract key features and form training resources for accident tripping analysis;

[0013] S3. Construction and application of accident tripping diagnosis and analysis model: The diagnostic analysis model is constructed using machine learning algorithms. The model is trained using historical tripping data. When an accident tripping occurs, real-time multi-source fusion data is input into the model to automatically identify the cause of the tripping and the location of the fault and provide handling suggestions.

[0014] S4. Automatic generation of accident trip reports: Construct a rule model for automatic report generation, develop standardized templates based on production operation data and business management specifications, automatically fill in the report content according to the diagnostic results, and support multi-dimensional query and export;

[0015] S5. Diagnostic Result Verification and Model Optimization: The diagnostic results are compared and verified with the actual maintenance results, and the fusion model and diagnostic analysis model are iteratively optimized using the verification data.

[0016] According to the above technical solution, in the multi-source data acquisition and preprocessing in step S1, time series alignment of data from different sources is performed through a unified data interface standard, and outlier detection algorithm is used to remove noisy data.

[0017] According to the above technical solution, the feature extraction and dimensionality reduction algorithm in step S2 includes principal component analysis or deep autoencoder, and the machine learning algorithm includes random forest, neural network or gradient boosting tree.

[0018] According to the above technical solution, the accident trip report in step S3 includes the trip time, cause of the fault, scope of impact, and handling suggestions.

[0019] According to the above technical solution, in the diagnostic result verification and model optimization in step S5, verification data is collected by comparing the diagnostic results with the actual maintenance results, and the model parameters are iteratively updated using supervised learning.

[0020] A digital-scenario substation operation accident tripping diagnosis system includes:

[0021] Data acquisition module: used to collect telemetry and telecontrol parameters, meteorological information, defect records, video surveillance data and fault recording files related to substation accident tripping in real time, and communicate with the substation smart gateway through a standardized interface;

[0022] Data preprocessing module: Cleans, classifies, integrates and correlates the collected data, including data cleaning subunit, data integration subunit and feature extraction subunit;

[0023] The diagnostic model module includes a multi-source data fusion model, a tripping accident diagnostic analysis model, and a report generation rule model. The multi-source data fusion model performs feature fusion on preprocessed data, the diagnostic analysis model identifies the cause of tripping based on machine learning algorithms, and the report generation rule model generates standardized reports.

[0024] Diagnostic execution module: Receives trip event trigger signals, calls the diagnostic model module to analyze real-time data, outputs fault cause, location and handling suggestions, and triggers the report generation process;

[0025] Report generation module: Automatically generates accident trip briefings and quick reports based on the report generation rule model, and supports report storage, querying, exporting and message push;

[0026] Model optimization module: Collects comparison data between diagnostic results and actual maintenance results, and iteratively trains the diagnostic model module.

[0027] According to the above technical solution, the data cleaning subunit adopts an outlier detection and repair algorithm, the data integration subunit realizes time alignment and format unification of multi-source data, and the feature extraction subunit adopts a principal component analysis algorithm.

[0028] According to the above technical solution, in the diagnostic model module, the multi-source data fusion model uses a deep autoencoder to extract data features, the accident tripping diagnostic analysis model is constructed based on the gradient boosting tree algorithm, and the report generation rule model includes a standardized template library that conforms to business management specifications.

[0029] According to the above technical solution, the message push methods of the report generation module include SMS, Elink messages and system messages, and support the automatic push of accident tripping reports to relevant responsible persons.

[0030] According to the above technical solution, the model optimization module iteratively trains the diagnostic model module through supervised learning, and the optimization parameters include feature weights, model structure and classification threshold.

[0031] Compared with the prior art, the method and system for diagnosing tripping accidents in substations based on digital scenarios provided by the present invention have at least the following beneficial effects:

[0032] (1) This invention achieves automatic identification of the cause of accident tripping through multi-source data fusion and intelligent algorithms. Compared with traditional manual analysis, the fault location time is shortened from hours to minutes, the diagnostic accuracy is increased to more than 90%, and human error is reduced. In addition, it automatically generates accident tripping reports that meet business specifications, and the report generation time is shortened from several hours to minutes, providing timely decision support for fault handling and reducing the risk of accident expansion.

[0033] (2) This invention integrates multi-dimensional operational data to build a data-driven diagnostic model, realizing the transformation from "experience-driven" to "data-driven" diagnostic mode, providing a scientific basis for substation operation and maintenance; automated diagnosis and report generation reduce the workload of operation and maintenance personnel, freeing up human resources, allowing them to devote more energy to fault handling and system optimization, and improving overall operation and maintenance efficiency.

[0034] (3) Through iterative optimization of machine learning algorithms, the system can continuously learn new fault modes, adapt to the complex operating environment of substations, reduce reliance on human experience, promote the development of power system operation and maintenance towards intelligence and unmanned operation, and realize full-process digital management from data collection, diagnosis and analysis to report generation by combining digital scenarios. This provides core technical support for the panoramic and intelligent management of substations and helps the construction of digital power grid. Attached Figure Description

[0035] Figure 1 This is a flowchart of the method for diagnosing tripping accidents in substations based on digital scenarios, as described in this invention.

[0036] Figure 2 This is a structural diagram of the digital scenario-based substation production and operation accident tripping diagnosis system of the present invention. Detailed Implementation

[0037] The present invention will be further described below with reference to embodiments.

[0038] Please see Figure 1 This invention provides a method for diagnosing tripping accidents in substations based on digital scenarios, comprising the following steps:

[0039] Step S1, multi-source data acquisition and preprocessing, specifically includes:

[0040] Real-time acquisition of multi-source heterogeneous data, including equipment telemetry and telecommunication parameters, meteorological information, defect records, alarm notifications, on-site video monitoring data, information protection substation data, and fault recording files, before and after a substation accident trip.

[0041] The collected data is preprocessed, including data cleaning (removing noise and outliers), classification and summarization, data integration and correlation mapping, to build a standardized accident trip data pool; for example, by using a unified data interface standard, the equipment status data from different sources are aligned with the time series to eliminate data format differences.

[0042] Step S2, Multi-source data fusion model construction, specifically:

[0043] Based on the preprocessed data, a multi-source data preprocessing fusion model is constructed. Feature extraction and dimensionality reduction algorithms, such as principal component analysis (PCA) or deep autoencoders, are used to extract key features from the data, forming training resources for fault trip analysis. For example, abrupt changes in current and voltage are extracted from telemetry data, and abnormal weather factors such as lightning strikes and strong winds are correlated from meteorological data.

[0044] Step S3, Construction and Application of Fault Trip Diagnostic Analysis Model, specifically:

[0045] A fault trip diagnosis and analysis model is built using machine learning algorithms (such as random forest, neural network or gradient boosting tree). The model is trained using historical trip data to learn the characteristic patterns of different fault types.

[0046] When a fault trip occurs, the multi-source fusion data collected in real time is input into the diagnostic model. The model automatically identifies the cause of the trip and the location of the fault, and provides handling suggestions. For example, the model can determine whether it is a line short circuit, transformer fault or other type of fault based on the waveform distortion characteristics in the fault recording data and the action information of the protection device.

[0047] Step S4: Automatic generation of the fault trip report, specifically:

[0048] Develop an automatic generation rule model for accident trip reports, and formulate standardized report templates based on production operation data and business management specifications.

[0049] Based on the output of the diagnostic model, the system automatically populates the report content, including tripping time, fault cause, scope of impact, and handling suggestions, and supports multi-dimensional querying and export of the report. For example, the system can automatically generate accident briefings and quick reports that meet the requirements of the "Regulations for Investigation of Power System Accidents" based on the diagnostic results.

[0050] Step S5, diagnostic result verification and model optimization, specifically includes:

[0051] The automatically generated diagnostic results are compared and verified with the actual inspection results, and verification data is collected.

[0052] The multi-source data fusion model and diagnostic analysis model are iteratively optimized using validation data to improve the diagnostic accuracy and generalization ability of the models.

[0053] Please see Figure 2 This invention provides a digital-scenario substation operation accident tripping diagnosis system, comprising:

[0054] Data acquisition module: Used to collect multi-source heterogeneous data related to substation accident tripping in real time, including telemetry and teleindication unit, meteorological information unit, defect recording unit, video monitoring unit and fault recording unit, etc. Each unit communicates with the substation smart gateway through standardized interface to realize real-time data acquisition.

[0055] Data preprocessing module: Cleans, classifies, integrates and correlates the collected data to build an accident trip data pool; this module includes a data cleaning subunit (performs outlier detection and repair), a data integration subunit (achieves time alignment and format unification of multi-source data), and a feature extraction subunit (extracts key features using algorithms such as PCA).

[0056] The diagnostic model module includes a multi-source data fusion model, a tripping accident diagnostic analysis model, and a report generation rule model. The multi-source data fusion model performs feature fusion on preprocessed data; the diagnostic analysis model uses machine learning algorithms to identify the cause of the tripping accident; and the report generation rule model automatically generates standardized reports based on the diagnostic results.

[0057] Diagnostic execution module: Receives trip event trigger signals, calls the diagnostic model module to analyze real-time data, outputs fault causes, locations and handling suggestions, and triggers the report generation process.

[0058] Report generation module: Based on the report generation rule model, automatically generate accident trip briefings and quick reports, and support the storage, query, export and message push of reports (such as SMS and Elink message notifications to relevant responsible persons).

[0059] Model optimization module: Collects comparative data between diagnostic results and actual maintenance results, iteratively trains the diagnostic model module, optimizes model parameters, and improves diagnostic accuracy.

[0060] This solution has the following working process: It adopts a closed-loop working mode of "data acquisition - fusion analysis - intelligent diagnosis - report generation - model optimization". Through the collaborative operation of multiple modules, it realizes the intelligent processing of the entire process of substation accident tripping. When a substation accident trips, the system automatically triggers the diagnosis process, extracts key information from multi-source heterogeneous data, quickly locates the cause of the fault through intelligent algorithms, automatically generates a standardized diagnosis report, and iteratively optimizes the diagnosis model based on actual maintenance results to form a continuously evolving intelligent diagnosis system.

[0061] In summary: By integrating multi-source data and intelligent algorithms, the system achieves automatic identification of the causes of power outages. Compared to traditional manual analysis, fault location time is reduced from hours to minutes, and diagnostic accuracy is increased to over 90%, reducing human error. Furthermore, it automatically generates outage reports that conform to business specifications, reducing report generation time from hours to minutes, providing timely decision support for fault handling and reducing the risk of accident escalation. By integrating multi-dimensional operational data, a data-driven diagnostic model is constructed, shifting the diagnostic mode from "experience-driven" to "data-driven," providing a scientific basis for substation operation and maintenance. Automated diagnosis and report generation reduce the workload of operation and maintenance personnel, freeing up human resources to focus more on fault handling and system optimization, improving overall operation and maintenance efficiency. Through iterative optimization of machine learning algorithms, the system can continuously learn new fault modes, adapt to the complex operating environment of substations, reduce reliance on human experience, and promote the development of power system operation and maintenance towards intelligence and unmanned operation. Combined with digital scenarios, it achieves full-process digital management from data collection and diagnostic analysis to report generation, providing core technical support for panoramic and intelligent substation management and contributing to the construction of a digital power grid.

[0062] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0063] In conclusion, 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 digitalized scene-based substation production operation fault tripping diagnosis method, characterized in that: Includes the following steps: S1. Multi-source data acquisition and preprocessing: Real-time acquisition of equipment telemetry and teleindication parameters, meteorological information, defect records, alarm notifications, on-site video monitoring data, information protection substation data and fault recording files before and after substation accident tripping; cleaning, classifying and summarizing the acquired data, integrating the data and mapping the associations to build a standardized accident tripping data pool. S2. Construction of multi-source data fusion model: Based on the preprocessed data, feature extraction and dimensionality reduction algorithms are used to extract key features and form training resources for accident tripping analysis; S3. Construction and application of accident tripping diagnosis and analysis model: The diagnostic analysis model is constructed using machine learning algorithms. The model is trained using historical tripping data. When an accident tripping occurs, real-time multi-source fusion data is input into the model to automatically identify the cause of the tripping and the location of the fault and provide handling suggestions. S4. Automatic generation of accident trip reports: Construct a rule model for automatic report generation, develop standardized templates based on production operation data and business management specifications, automatically fill in the report content according to the diagnostic results, and support multi-dimensional query and export; S5. Diagnostic Result Verification and Model Optimization: The diagnostic results are compared and verified with the actual maintenance results, and the fusion model and diagnostic analysis model are iteratively optimized using the verification data.

2. The digitalized scene-based substation production operation fault tripping diagnosis method according to claim 1, characterized in that: In step S1, the multi-source data acquisition and preprocessing process involves aligning data from different sources to time series using a unified data interface standard and removing noisy data using an outlier detection algorithm.

3. The digitalized scene-based substation production operation fault tripping diagnosis method according to claim 2, characterized in that: The feature extraction and dimensionality reduction algorithm in step S2 includes principal component analysis or deep autoencoder, and the machine learning algorithm includes random forest, neural network or gradient boosting tree.

4. The method for diagnosing tripping accidents in substations based on digital scenarios according to claim 3, characterized in that: The trip report in step S3 includes the trip time, cause of the fault, scope of impact, and handling recommendations.

5. The method for diagnosing tripping accidents in substations based on digital scenarios according to claim 4, characterized in that: In step S5, during the diagnostic result verification and model optimization, verification data is collected by comparing the diagnostic results with the actual maintenance results, and the model parameters are iteratively updated using supervised learning.

6. A digital-scenario-based substation operation accident tripping diagnosis system, characterized in that: include: Data acquisition module: used to collect telemetry and telecontrol parameters, meteorological information, defect records, video surveillance data and fault recording files related to substation accident tripping in real time, and communicate with the substation smart gateway through a standardized interface; Data preprocessing module: Cleans, classifies, integrates and correlates the collected data, including data cleaning subunit, data integration subunit and feature extraction subunit; The diagnostic model module includes a multi-source data fusion model, a tripping accident diagnostic analysis model, and a report generation rule model. The multi-source data fusion model performs feature fusion on preprocessed data, the diagnostic analysis model identifies the cause of tripping based on machine learning algorithms, and the report generation rule model generates standardized reports. Diagnostic execution module: Receives trip event trigger signals, calls the diagnostic model module to analyze real-time data, outputs fault cause, location and handling suggestions, and triggers the report generation process; Report generation module: Automatically generates accident trip briefings and quick reports based on the report generation rule model, and supports report storage, querying, exporting and message push; Model optimization module: Collects comparison data between diagnostic results and actual maintenance results, and iteratively trains the diagnostic model module.

7. The substation production and operation accident tripping diagnosis system based on digital scenario as described in claim 6, characterized in that: The data cleaning subunit employs an outlier detection and repair algorithm, the data integration subunit achieves time alignment and format unification of multi-source data, and the feature extraction subunit employs a principal component analysis algorithm.

8. The substation production and operation accident tripping diagnosis system based on digital scenario as described in claim 7, characterized in that: In the diagnostic model module, the multi-source data fusion model uses a deep autoencoder to extract data features, the accident tripping diagnostic analysis model is built based on the gradient boosting tree algorithm, and the report generation rule model includes a standardized template library that conforms to business management specifications.

9. The substation production and operation accident tripping diagnosis system based on digital scenario as described in claim 8, characterized in that: The report generation module can push messages via SMS, Elink messages, and system messages, and supports automatically pushing accident tripping reports to relevant responsible persons.

10. The substation production and operation accident tripping diagnosis system based on digital scenario as described in claim 9, characterized in that: The model optimization module iteratively trains the diagnostic model module through supervised learning, optimizing parameters including feature weights, model structure, and classification threshold.

Images