CN117674128A

CN117674128A - Automatic fault removal method based on power dispatching system

Info

Publication number: CN117674128A
Application number: CN202311683968.4A
Authority: CN
Inventors: 陈智歆; 钟昱炜; 周江华; 梁威魄; 汪志昆; 林习艺
Original assignee: Quanzhou Power Supply Co of State Grid Fujian Electric Power Co Ltd
Current assignee: Quanzhou Power Supply Co of State Grid Fujian Electric Power Co Ltd
Priority date: 2023-12-08
Filing date: 2023-12-08
Publication date: 2024-03-08

Abstract

The invention relates to the technical field of power system control methods, in particular to an automatic fault removal method based on a power dispatching system. According to the invention, through a time sequence analysis method, data acquisition and preliminary analysis are more accurate, a detailed preliminary power grid analysis report is generated, the recognition and prediction capability of a fault mode is enhanced by utilizing a convolutional neural network and a cyclic neural network algorithm, the accuracy and timeliness of fault prediction are improved, and an automatic alarm and processing scheme is formulated, so that fault response is rapid and efficient, manual intervention is reduced, a support vector machine algorithm provides accurate decision support in terms of fault isolation and processing, a genetic algorithm promotes efficient utilization in resource optimization scheduling, and a reinforcement learning and simulated annealing algorithm optimizes protection strategy and system performance, and the stability and reliability of a power scheduling system are enhanced.

Description

Automatic fault removal method based on power dispatching system

Technical Field

The invention relates to the technical field of power system control methods, in particular to an automatic fault removal method based on a power dispatching system.

Background

The technical field of power system control methods is focused on ensuring the stability, safety and efficiency of power supply. This area covers the whole process from power generation, transmission to distribution, with emphasis on monitoring grid status in real time, managing load scheduling, fault detection and isolation, system optimization, and safety precautions. The purpose of these measures is to ensure the continuity and quality of the power supply, while optimizing the use of resources, improving energy efficiency and economic efficiency.

An automatic fault removal method based on a power dispatching system is an important component in the field of power system control, and an automatic technology is utilized to identify and solve faults in a power system. The main purpose of the method is to improve the response speed and the processing efficiency of the power system to faults. The method is introduced to reduce the power failure time caused by faults, ensure the stable operation of the power system and reduce the risk of human operation errors. Finally, the method aims at improving the reliability of the power system, reducing the maintenance cost and improving the satisfaction degree of users on the power service. The implementation of automated troubleshooting methods based on power dispatching systems generally includes a series of comprehensive means, mainly including data analysis and monitoring, fault diagnosis algorithms, automatic control systems, and remote operations and interventions. These methods work together to collect and analyze grid data, such as voltage, current, and frequency, in real-time to monitor system status and predict potential faults in real-time. Once anomalies are detected, advanced algorithms are used to quickly and accurately locate the source and nature of the fault. The system automatically performs the necessary control operations, such as isolating the fault area, to minimize the impact of the fault on the overall power system. The remote operation and intervention function allows operation and maintenance personnel to remotely monitor the system, respond and process faults in time, and further improves the speed and efficiency of fault processing. The combined use of these approaches enables the power system to more intelligently and adaptively address various faults and challenges, thereby improving overall operational stability and efficiency.

The traditional power dispatching system automatic fault removal method has a plurality of defects. These methods often rely on simple fault detection and response mechanisms, lack deep data analysis and fault pattern recognition capabilities, resulting in lower fault prediction accuracy. In addition, traditional methods are generally passive in terms of fault handling policy formulation, rely on manual intervention, are inefficient, and are prone to error. In terms of resource scheduling and optimization, the traditional method lacks efficient algorithm support, so that resource utilization is not optimized enough, and complex and changeable power grid environments cannot be effectively treated. Finally, these approaches also exhibit deficiencies in long-term system performance enhancement and risk assessment, lacking effective strategies for overall optimization and performance enhancement of the system.

Disclosure of Invention

The invention aims to solve the defects existing in the prior art, and provides an automatic fault removal method based on a power dispatching system.

In order to achieve the above purpose, the present invention adopts the following technical scheme: the automatic fault removal method based on the power dispatching system comprises the following steps:

s1: based on the power grid node, adopting a time sequence analysis method to perform data acquisition and preliminary analysis, and generating a preliminary power grid analysis report;

S2: based on the preliminary power grid analysis report, performing fault mode identification by adopting a convolutional neural network algorithm, and generating a fault mode identification report;

s3: based on the fault mode identification report, performing fault prediction by adopting a cyclic neural network algorithm, and generating a fault prediction report;

s4: based on the fault prediction report, carrying out automatic alarm and suggestion formulation to generate an automatic alarm and processing scheme;

s5: based on the automatic alarm and processing scheme, adopting a support vector machine algorithm to perform fault isolation and processing, and generating fault isolation and recovery operation;

s6: based on the fault isolation and recovery operation, adopting a genetic algorithm to perform resource optimization scheduling, and generating a resource optimization scheduling plan;

s7: based on the resource optimization scheduling plan, strategy generation and risk assessment are carried out by adopting a reinforcement learning algorithm, and generation is carried out

Protection strategy and risk assessment;

s8: based on the protection strategy and the risk assessment, adopting a simulated annealing algorithm to perform system optimization and feedback, and generating system optimization and performance feedback.

As a further scheme of the invention, the preliminary power grid analysis report specifically comprises a preliminary analysis result of load change, meteorological conditions and historical fault record data, the fault mode identification report specifically comprises an identified abnormal mode and power grid trend, the fault prediction report comprises a predicted potential fault point and a predicted fault type, the fault isolation and recovery operation specifically comprises an operation flow of automatically isolating a fault area and starting a standby route, and the protection strategy and risk assessment comprises an assessment of protection strategies aiming at preset conditions and potential risks and benefits thereof.

As a further scheme of the invention, based on the power grid node, a time sequence analysis method is adopted to perform data acquisition and preliminary analysis, and a preliminary power grid analysis report is generated:

s101: based on the power grid node, collecting load change, meteorological conditions and historical fault record data by adopting a data collection technology, and generating an original data set;

s102: based on the original data set, adopting a data preprocessing technology to carry out data cleaning and formatting processing to generate a preprocessed data set;

s103: based on the preprocessed data set, performing preliminary data exploration by adopting a statistical analysis method, and generating a preliminary data exploration report;

s104: based on the preliminary data exploration report, adopting a time sequence analysis method to deeply analyze data trend, and generating a preliminary power grid analysis report;

the data collection technology specifically comprises sensor data acquisition and system log analysis, the data preprocessing technology comprises data denoising, normalization and data format conversion, the statistical analysis method specifically comprises exploratory data analysis and principal component analysis, and the time sequence analysis method comprises trend analysis and seasonal decomposition.

As a further scheme of the invention, based on the preliminary power grid analysis report, a convolutional neural network algorithm is adopted to perform fault mode identification, and a fault mode identification report is generated:

S201: based on the preliminary power grid analysis report, extracting key features by adopting a data feature extraction technology, and generating a feature extraction report;

s202: training a deep learning model by adopting a convolutional neural network algorithm based on the feature extraction report to generate a trained CNN model;

s203: based on the trained CNN model, carrying out deep analysis on the characteristic data, identifying potential fault modes, and generating a fault mode deep analysis result;

s204: based on the fault mode depth analysis result, comprehensively utilizing a model verification and result verification technology to judge the fault mode and generate a fault mode identification report;

the data feature extraction technique includes feature engineering and correlation analysis, the convolutional neural network algorithm is used for identifying data patterns, and the depth analysis includes depth pattern identification and trend prediction.

As a further scheme of the invention, based on the fault mode identification report, a cyclic neural network algorithm is adopted to conduct fault prediction, and a fault prediction report is generated:

s301: preprocessing data by adopting a data normalization and normalization technology based on the fault mode identification report to generate a standardized data set;

S302: based on the standardized data set, constructing an LSTM fault prediction model by adopting a long-short-term memory network algorithm;

s303: based on the LSTM fault prediction model, model training and super-parameter optimization are executed, prediction accuracy is improved, and an optimized LSTM model is generated;

s304: performing fault prediction based on the optimized LSTM model to generate a fault prediction report;

the data normalization technique includes outlier removal and data scaling, the long-short term memory network is used for feature learning and pattern recognition of time series data, the model training includes cross validation and loss function minimization, and the fault prediction includes detection of potential fault points and classification of fault types.

As a further scheme of the invention, based on the fault prediction report, automatic alarm and suggestion formulation is carried out, and an automatic alarm and processing scheme is generated:

s401: based on the fault prediction report, adopting a rule engine and a logic analysis technology to formulate a fault alarm rule and generate a fault alarm rule set;

s402: based on the fault alarm rule set, executing fault detection and alarm by using a real-time data monitoring system, and generating a real-time fault alarm notification;

S403: based on the real-time fault alarm notification, adopting an emergency response strategy formulation technology to draw fault coping measures and generating a fault coping strategy draft;

s404: based on the fault coping strategy draft, comprehensive examination and optimization are carried out, and an automatic alarm and processing scheme is generated;

the rule engine technology comprises condition judgment and threshold setting, the real-time data monitoring system comprises data flow processing and abnormal event triggering, the emergency response strategy formulation comprises risk assessment and prioritization, and the examination and optimization comprises actual feasibility assessment and resource allocation consideration of the strategy.

As a further scheme of the invention, based on the automatic alarm and processing scheme, a support vector machine algorithm is adopted to perform fault isolation and processing, and fault isolation and recovery operation is generated:

s501: based on the automatic alarm and processing scheme, adopting a data analysis technology to identify key fault characteristics and generating a fault characteristic data set;

s502: based on the fault characteristic data set, a support vector machine algorithm is adopted to generate a fault isolation model;

s503: based on the fault isolation model, model training and adjustment are carried out, fault isolation precision is guaranteed, and an adjusted fault isolation model is generated;

S504: based on the adjusted fault isolation model, fault isolation and recovery operations are implemented;

the support vector machine algorithm is used for classification and anomaly detection, the model adjustment comprises parameter optimization and verification, and the fault isolation and recovery operation comprises the identification of a fault area and the implementation of isolation measures.

As a further scheme of the invention, based on the fault isolation and recovery operation, a genetic algorithm is adopted to carry out resource optimization scheduling, and a resource optimization scheduling plan is generated:

s601: based on the fault isolation and recovery operation, adopting a resource state analysis technology to evaluate resource allocation requirements and generating a resource allocation requirement analysis report;

s602: generating a resource scheduling optimization model by adopting a genetic algorithm based on the resource allocation demand analysis report;

s603: based on the resource scheduling optimization model, model training and optimization are executed, resource allocation efficiency is ensured, and an optimized resource scheduling model is generated;

s604: based on the optimized resource scheduling model, implementing resource optimization scheduling to generate a resource optimization scheduling plan;

the resource state analysis technology comprises resource utilization rate evaluation and demand prediction, the genetic algorithm is used for finding out an optimal resource allocation scheme, the model training and optimization comprise fitness evaluation and population evolution strategies, and the scheduling plan comprises resource reallocation and priority management.

As a further scheme of the invention, based on the resource optimization scheduling plan, a reinforcement learning algorithm is adopted to perform policy generation and risk assessment, and a protection policy and risk assessment are generated:

s701: based on the resource optimization scheduling plan, evaluating the current resource scheduling environment by adopting an environment analysis technology, and generating a resource scheduling environment analysis report;

s702: simulating a differentiated resource scheduling strategy by adopting a reinforcement learning algorithm based on the resource scheduling environment analysis report to generate a strategy simulation result;

s703: based on the strategy simulation result, carrying out strategy effect evaluation and risk analysis to generate a strategy effect evaluation report;

s704: comprehensively evaluating multiple strategies and selecting an optimal strategy based on the strategy effect evaluation report, and generating a protection strategy and risk evaluation;

the environmental analysis techniques include resource availability and demand prediction, the reinforcement learning algorithm is used for exploration and utilization of strategies, the evaluation includes benefit analysis and potential risk identification, and the comprehensive evaluation includes cost benefit analysis and risk management.

As a further scheme of the invention, based on the protection strategy and risk assessment, a simulated annealing algorithm is adopted to perform system optimization and feedback, and system optimization and performance feedback are generated:

S801: based on the protection strategy and the risk assessment, analyzing the current system performance by adopting a performance monitoring technology, and generating a system performance analysis report;

s802: based on the system performance analysis report, adopting a simulated annealing algorithm to optimize system configuration and performance and generating a system optimization scheme;

s803: based on the system optimization scheme, executing optimization operation and adjustment, maintaining the optimal performance of an operating system, and generating an execution result of the optimization operation;

s804: based on the execution result of the optimization operation, performing system performance evaluation and feedback to generate system optimization and performance feedback;

the performance monitoring technology comprises system running state monitoring and efficiency evaluation, the simulated annealing algorithm is used for global optimization, the optimization operation comprises parameter adjustment and function enhancement, and the performance evaluation and feedback comprises comprehensive evaluation of system stability and efficiency.

Compared with the prior art, the invention has the advantages and positive effects that:

according to the invention, by using a time sequence analysis method, data acquisition and preliminary analysis are more accurate, and a finer preliminary power grid analysis report can be generated. The convolutional neural network and the cyclic neural network algorithm are utilized, so that the recognition and prediction capability of the fault mode is enhanced, and the accuracy and timeliness of fault prediction are improved. And the automatic alarm and processing scheme is formulated, so that the fault response is quicker and more efficient, and the need of manual intervention is reduced. The application of support vector machine algorithms provides more accurate decision support in terms of fault isolation and handling, while genetic algorithms promote efficient utilization of resources in resource-optimized scheduling. Through reinforcement learning and simulated annealing algorithm, the protection strategy and system performance are continuously optimized and improved, and the stability and reliability of the whole power dispatching system are enhanced.

Drawings

FIG. 1 is a schematic diagram of the main steps of the present invention;

FIG. 2 is a detailed schematic of the S1 of the present invention;

FIG. 3 is a schematic diagram of an S2 refinement of the present invention;

FIG. 4 is a schematic diagram of an S3 refinement of the present invention;

FIG. 5 is a schematic diagram of an S4 refinement of the present invention;

FIG. 6 is a schematic diagram of an S5 refinement of the present invention;

FIG. 7 is a schematic diagram of an S6 refinement of the present invention;

FIG. 8 is a schematic diagram of an S7 refinement of the present invention;

fig. 9 is a schematic diagram of the S8 refinement of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

In the description of the present invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like indicate orientations or positional relationships based on the orientation or positional relationships shown in the drawings, merely to facilitate describing the present invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present invention. Furthermore, in the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.

Example 1

Referring to fig. 1, the present invention provides a technical solution: the automatic fault removal method based on the power dispatching system comprises the following steps:

s1: based on the power grid nodes, adopting a time sequence analysis method to perform data acquisition and preliminary analysis to generate preliminary power grid components

Analyzing a report;

s5: based on an automatic alarm and processing scheme, performing fault isolation and processing by adopting a support vector machine algorithm, and generating fault isolation and recovery operation;

s6: based on fault isolation and recovery operation, adopting a genetic algorithm to perform resource optimization scheduling, and generating a resource optimization scheduling plan;

s7: based on the resource optimization scheduling plan, performing strategy generation and risk assessment by adopting a reinforcement learning algorithm, and generating a protection strategy and risk assessment;

The preliminary power grid analysis report specifically comprises a load change, meteorological conditions and preliminary analysis results of historical fault record data, the fault mode identification report specifically comprises an identified abnormal mode and a power grid trend, the fault prediction report comprises a predicted potential fault point and a predicted fault type, the fault isolation and recovery operation specifically comprises an operation flow of automatically isolating a fault area and starting a standby route, and the protection strategy and risk assessment comprises an assessment of protection strategy aiming at preset conditions and potential risks and benefits thereof.

By means of a time sequence analysis method, the system can monitor the state of the power grid in real time and forecast future trends, so that potential problems can be dealt with in advance. The convolutional neural network and the cyclic neural network are applied to enhance the accuracy and the prediction capability of fault diagnosis, so that operation and maintenance personnel can respond to faults quickly and effectively. The generation of an automatic alarm and processing scheme and the application of a support vector machine algorithm in fault isolation and processing improve the speed and efficiency of emergency response and ensure the rapid processing when faults occur. The use of genetic algorithms in resource optimization scheduling improves the utilization efficiency of resources, while the use of reinforcement learning and simulated annealing algorithms further enhances the adaptive capacity and optimization performance of the system.

Referring to fig. 2, based on the grid node, a time series analysis method is adopted to perform data acquisition and preliminary analysis, and a preliminary grid analysis report is generated:

In S101, real-time data such as load changes and meteorological conditions of the grid nodes are collected by deploying high-precision sensors. System log analysis is used to extract historical fault records and operational data. These data together form the original dataset, providing the basis for in-depth analysis.

In S102, the data cleansing technique is used to remove extraneous or erroneous data records and normalize the data range by normalization processing. Data format conversion is also required to ensure that the data is compatible for use in subsequent analysis.

In S103, statistical methods such as exploratory data analysis and principal component analysis are used. The goal of this stage is to identify and understand key trends, patterns and potential outliers in the data, which conveniently lay the foundation for deeper analysis.

In S104, based on the findings of the previous steps, the time-series characteristics of the data, including long-term trends and seasonal patterns, are deeply analyzed. The key point of the step is to identify and interpret time-related factors affecting the performance of the power grid, and provide basis for developing improvement measures and future planning.

Referring to fig. 3, based on the preliminary grid analysis report, a convolutional neural network algorithm is adopted to perform fault mode identification, and a fault mode identification report is generated:

the data feature extraction technology comprises feature engineering and correlation analysis, a convolutional neural network algorithm is used for identifying data patterns, and depth analysis comprises depth pattern identification and trend prediction.

In S201, important features are screened from the original data by using feature engineering and correlation analysis data feature extraction technology, and these features can effectively reflect the operation condition of the power grid.

In S202, based on the result in the feature extraction report, a deep learning model is constructed using a convolutional neural network algorithm, and the extracted features are trained.

In S203, a test is performed on the new power grid state data by using the trained CNN model, and the feature data is subjected to a deep analysis to identify a potential failure mode.

In S204, based on the fault mode deep analysis result, the power grid fault mode is accurately determined through the model verification and the result verification technology.

Referring to fig. 4, based on the failure mode identification report, a cyclic neural network algorithm is adopted to perform failure prediction, and a failure prediction report is generated:

s301: preprocessing data by adopting a data standardization and normalization technology based on a fault mode identification report to generate a standardized data set;

s302: based on a standardized data set, constructing an LSTM fault prediction model by adopting a long-short-term memory network algorithm;

the data normalization technology comprises outlier removal and data scale transformation, the long-short-term memory network is used for feature learning and pattern recognition of time series data, the model training comprises cross validation and loss function minimization, and the fault prediction comprises detection of potential fault points and classification of fault types.

In S301, normalization and normalization processing are performed using data in the failure mode identification report. This step includes removing outliers and performing a data scaling to ensure the quality of the data input to the LSTM model. The processed data set will be standardized for ease of processing in subsequent steps.

In S302, an LSTM failure prediction model is constructed based on the normalized dataset. LSTM networks are particularly suited for processing time series data, effectively capturing long-term dependencies in the data. The built model is dedicated to learning the time series characteristics of the grid failure mode and to identifying potential failure modes.

In S303, training of the LSTM model and super-parametric optimization are performed. This process includes using cross-validation to prevent overfitting and adjusting the loss function to minimize the prediction error. The method aims to improve the prediction accuracy of the model on actual power grid data. The trained and optimized LSTM model will be used for fault prediction.

In S304, fault prediction is performed using the optimized LSTM model. This step includes the detection of potential failure points and the classification of failure types. The prediction result is recorded in the fault prediction report in detail, so that key information is provided for a power grid operation and maintenance team, preventive measures are conveniently taken in advance, and the influence of faults is avoided or reduced.

Referring to fig. 5, automatic alarm and advice formulation is performed based on the fault prediction report, and an automatic alarm and processing scheme is generated:

the rule engine technology comprises condition judgment and threshold setting, the real-time data monitoring system comprises data flow processing and abnormal event triggering, emergency response strategy formulation comprises risk assessment and prioritization, and examination and optimization comprise actual feasibility assessment and resource allocation consideration of the strategy.

In S401, a proper rule engine technique is selected, and condition judgment and threshold setting are set. The setting of these rule engines requires that critical information in the fault prediction report be fully considered to ensure that the rules accurately capture potential fault patterns. The rule set should be flexible to accommodate changes in the system operation.

In S402, the operation of the real-time data monitoring system is involved. Integrating data flow processing and an abnormal event triggering mechanism, and monitoring the running state of the system in real time. By means of the formulated fault alarm rule set, the system accurately identifies and alarms abnormal conditions and generates timely fault alarm notification.

In S403, risk assessment and prioritization are performed using emergency response policy formulation techniques. The emergency degree of the countermeasures is determined by considering the influence degree of different fault types on the system, and a corresponding emergency response flow is formulated, so that rapid and effective countermeasures are ensured when faults occur.

In S404, the formulated failure handling policy draft needs to be comprehensively reviewed. The performance of the strategy in the actual environment is evaluated through simulation test, the resource utilization efficiency is considered, and the implementation cost is reduced by optimizing the strategy. And determining the form and the receiving party of the alarm notification, and ensuring timely information transmission.

Referring to fig. 6, based on the automatic alarm and processing scheme, a support vector machine algorithm is adopted to perform fault isolation and processing, and a fault isolation and recovery operation is generated:

s501: based on an automatic alarm and processing scheme, adopting a data analysis technology to identify key fault characteristics and generating a fault characteristic data set;

s502: based on the fault characteristic data set, generating a fault isolation model by adopting a support vector machine algorithm;

the support vector machine algorithm is used for classification and anomaly detection, model adjustment comprises parameter optimization and verification, and fault isolation and recovery operation comprises fault region identification and isolation measure implementation.

In S501, the task: the key fault signature is identified using data analysis techniques.

The operation is as follows: raw data is collected, pre-processed (e.g., missing value processing, normalization, etc.), and then feature engineering techniques (e.g., principal component analysis, feature selection algorithms, etc.) are used to identify and extract key fault features.

Code example (assuming Python is used):

import pandas as pd

from sklearn.preprocessing import StandardScaler

from sklearn.decomposition import PCA

# load data

data＝pd.read_csv('fault_data.csv')

Pretreatment of #

data＝data.fillna(method＝'ffill')

scaler＝StandardScaler()

scaled_data＝scaler.fit_transform(data)

Feature extraction

pca＝PCA(n_components＝5)

feature_dataset＝pca.fit_transform(scaled_data)

In S502, the task: and constructing a fault isolation model by using a support vector machine algorithm.

The operation is as follows: appropriate SVM parameters (e.g., kernel function, C value, etc.) are selected and then the SVM model is trained with the failure feature dataset.

Code example:

from sklearn.svm import SVC

# assume feature_dataset and labels are ready

model＝SVC(kernel＝'rbf',C＝1.0)

model.fit(feature_dataset,labels)

In S503, the task: model training and adjustment are carried out, and fault isolation precision is ensured.

The operation is as follows: cross-validation and grid search techniques are used to optimize the model parameters.

Code example:

in S504, the task: and performing fault isolation and recovery operation by using the adjusted model.

The operation is as follows: and predicting new data by using the trained model, identifying a fault area and implementing isolation measures.

Code example:

# assume new_data is new observation data

scaled_new_data＝scaler.transform(new_data)

fault_prediction＝optimized_model.predict(scaled_new_data)

# implement isolation measures based on prediction results

if fault_prediction＝＝[some_fault_condition]:

# implement isolation measures

Referring to fig. 7, a genetic algorithm is adopted to perform resource optimization scheduling based on fault isolation and recovery operations, and a resource optimization scheduling plan is generated:

s601: based on fault isolation and recovery operation, adopting a resource state analysis technology to evaluate resource allocation requirements and generating a resource allocation requirement analysis report;

the resource state analysis technology comprises resource utilization rate evaluation and demand prediction, the genetic algorithm is used for finding out an optimal resource allocation scheme, model training and optimization comprise fitness evaluation and population evolution strategies, and the scheduling plan comprises resource reallocation and priority management.

In S601, by applying the resource status analysis technique, the system evaluates the resource utilization and predicts the demand, and establishes the specific resource demand of the system in the fault isolation and recovery operation. This provides an explicit basis for the subsequent optimal scheduling.

In S602, the resource allocation requirement analysis report is converted into an optimizable model by using a genetic algorithm. And determining an optimization target and constraint conditions, abstracting the optimization target and constraint conditions into a problem form which can be processed by a genetic algorithm, and finding an optimal resource allocation scheme.

In S603, by using fitness evaluation and population evolution strategies, it is ensured that the model can effectively adapt to system changes, and resource allocation efficiency is improved. This step aims at generating an optimized resource scheduling model reflecting the optimal resource allocation strategy.

In S604, optimized scheduling of resources is performed based on the optimized resource scheduling model. Through resource reallocation and priority management, the system is ensured to maximally utilize available resources and minimize influence when in fault recovery. A detailed resource-optimized scheduling plan is generated, including a specific resource allocation scheme and execution timing.

Referring to fig. 8, policy generation and risk assessment are performed by adopting a reinforcement learning algorithm based on a resource optimization scheduling plan, and a protection policy and risk assessment are generated:

s702: based on the resource scheduling environment analysis report, adopting a reinforcement learning algorithm to simulate a differentiated resource scheduling strategy and generating a strategy simulation result;

environmental analysis techniques include resource availability and demand prediction, reinforcement learning algorithms for exploration and utilization of strategies, evaluation including benefit analysis and potential risk identification, and comprehensive evaluation including cost benefit analysis and risk management.

In S701, the system performs a comprehensive evaluation on the current resource scheduling environment, including consideration of resource availability and demand prediction, by using an environment analysis technique. The generated environmental analysis report provides a basis for the application of a subsequent reinforcement learning algorithm, and ensures that policy generation is performed on the basis of deep understanding of the system environment.

In S702, the reinforcement learning algorithm is used to simulate differentiated resource scheduling strategies, so that it is convenient to find an optimal scheduling scheme best adapted to the current environment by exploring and utilizing different strategies in the simulation. This ensures that the system is able to flexibly adjust the resource allocation policy according to actual environmental changes.

In S703, a policy effect evaluation report is generated by evaluating the result of the policy simulation, including benefit analysis and identification of potential risks. This report details the actual performance, advantages and disadvantages of the various strategies and the potential risk factors introduced.

In S704, based on the policy effect evaluation report, the system comprehensively evaluates a plurality of policies and selects an optimal policy. In the process, cost benefit analysis and risk management are carried out, so that the reliability and benefit of resource scheduling can be ensured in practical application by the finally generated protection strategy and risk assessment.

Referring to fig. 9, based on protection policy and risk assessment, a simulated annealing algorithm is used to perform system optimization and feedback, and generate system optimization and performance feedback:

s801: based on protection strategy and risk assessment, adopting a performance monitoring technology to analyze the current system performance and generate a system performance analysis report;

s802: based on a system performance analysis report, adopting a simulated annealing algorithm to optimize system configuration and performance and generating a system optimization scheme;

s803: based on a system optimization scheme, executing optimization operation and adjustment, maintaining the optimal performance of an operating system, and generating an optimization operation execution result;

the performance monitoring technology comprises system running state monitoring and efficiency evaluation, a simulated annealing algorithm is used for global optimization, the optimization operation comprises parameter adjustment and function enhancement, and the performance evaluation and feedback comprise comprehensive evaluation of system stability and efficiency.

In S801, the current system is comprehensively analyzed by a performance monitoring technique. This includes real-time monitoring of the system operating conditions and efficiency assessment. The monitoring results are used to generate a system performance analysis report detailing the performance status of the system and the potential problems that exist.

In S802, according to a system performance analysis report, introducing a simulated annealing algorithm to perform global optimization. In this step, the simulated annealing algorithm searches the solution space for the optimal system configuration and performance tuning scheme. The generated system optimization scheme contains key information of parameter adjustment and performance optimization.

In S803, the system optimization scheme is applied to the actual system. This involves performing optimization operations and adjustments to ensure that the system is operating at optimal performance. This includes dynamic adjustment of parameters and enhancement of functionality to accommodate different workloads and environmental changes. And generating an execution result of the optimization operation after execution, and recording the performance and the change of the system in the actual application.

In S804, performance evaluation and feedback are performed on the system. This step comprehensively considers the stability and efficiency of the system and generates

System optimization and performance feedback. The feedback result is used for adjusting the protection strategy, re-evaluating the risk and feeding back to the whole system optimization flow to form a closed loop feedback system, so that the system can maintain the optimal performance in the continuously changing environment.

The present invention is not limited to the above embodiments, and any equivalent embodiments which can be changed or modified by the technical disclosure described above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above embodiments according to the technical matter of the present invention will still fall within the scope of the technical disclosure.

Claims

1. The automatic fault removal method based on the power dispatching system is characterized by comprising the following steps of:

based on the power grid node, adopting a time sequence analysis method to perform data acquisition and preliminary analysis, and generating a preliminary power grid analysis report;

based on the preliminary power grid analysis report, performing fault mode identification by adopting a convolutional neural network algorithm, and generating a fault mode identification report;

Based on the fault mode identification report, performing fault prediction by adopting a cyclic neural network algorithm, and generating a fault prediction report;

based on the fault prediction report, carrying out automatic alarm and suggestion formulation to generate an automatic alarm and processing scheme;

based on the automatic alarm and processing scheme, adopting a support vector machine algorithm to perform fault isolation and processing, and generating fault isolation and recovery operation;

based on the fault isolation and recovery operation, adopting a genetic algorithm to perform resource optimization scheduling, and generating a resource optimization scheduling plan;

based on the resource optimization scheduling plan, performing strategy generation and risk assessment by adopting a reinforcement learning algorithm, and generating a protection strategy and risk assessment;

based on the protection strategy and the risk assessment, adopting a simulated annealing algorithm to perform system optimization and feedback, and generating system optimization and performance feedback.

2. The automated power dispatching system-based troubleshooting method of claim 1, wherein the preliminary grid analysis report specifically includes a preliminary analysis result of load change, weather conditions, historical fault record data, the fault pattern recognition report specifically includes a recognized abnormal pattern and grid trend, the fault prediction report includes a predicted potential fault point and a predicted fault type, the fault isolation and restoration operation specifically refers to an operation flow of automatically isolating a fault area and starting a standby route, and the protection strategy and risk assessment includes an assessment of protection strategies for preset conditions and potential risks and benefits thereof.

3. The automatic fault removal method based on the power dispatching system according to claim 1, wherein based on the power grid nodes, a time sequence analysis method is adopted to perform data acquisition and preliminary analysis, and a preliminary power grid analysis report is generated:

based on the power grid node, collecting load change, meteorological conditions and historical fault record data by adopting a data collection technology, and generating an original data set;

based on the original data set, adopting a data preprocessing technology to carry out data cleaning and formatting processing to generate a preprocessed data set;

based on the preprocessed data set, performing preliminary data exploration by adopting a statistical analysis method, and generating a preliminary data exploration report;

based on the preliminary data exploration report, adopting a time sequence analysis method to deeply analyze data trend, and generating a preliminary power grid analysis report;

4. The automated power dispatching system-based troubleshooting method of claim 1 wherein based on the preliminary grid analysis report, a convolutional neural network algorithm is employed to perform fault pattern recognition and generate a fault pattern recognition report:

based on the preliminary power grid analysis report, extracting key features by adopting a data feature extraction technology, and generating a feature extraction report;

training a deep learning model by adopting a convolutional neural network algorithm based on the feature extraction report to generate a trained CNN model;

based on the trained CNN model, carrying out deep analysis on the characteristic data, identifying potential fault modes, and generating a fault mode deep analysis result;

based on the fault mode depth analysis result, comprehensively utilizing a model verification and result verification technology to judge the fault mode and generate a fault mode identification report;

5. The power dispatching system-based automated troubleshooting method of claim 1, wherein based on the failure mode identification report, a recurrent neural network algorithm is employed to conduct a failure prediction and generate a failure prediction report:

Preprocessing data by adopting a data normalization and normalization technology based on the fault mode identification report to generate a standardized data set;

based on the standardized data set, constructing an LSTM fault prediction model by adopting a long-short-term memory network algorithm;

based on the LSTM fault prediction model, model training and super-parameter optimization are executed, prediction accuracy is improved, and an optimized LSTM model is generated;

performing fault prediction based on the optimized LSTM model to generate a fault prediction report;

6. The automated power dispatching system-based troubleshooting method of claim 1 wherein automatic alarm and advice formulation is performed based on a fault prediction report, generating an automatic alarm and handling scheme:

based on the fault prediction report, adopting a rule engine and a logic analysis technology to formulate a fault alarm rule and generate a fault alarm rule set;

Based on the fault alarm rule set, executing fault detection and alarm by using a real-time data monitoring system, and generating a real-time fault alarm notification;

based on the real-time fault alarm notification, adopting an emergency response strategy formulation technology to draw fault coping measures and generating a fault coping strategy draft;

based on the fault coping strategy draft, comprehensive examination and optimization are carried out, and an automatic alarm and processing scheme is generated;

7. The automated power dispatching system-based fault clearing method of claim 1, wherein based on the automated alarm and handling scheme, a support vector machine algorithm is employed to perform fault isolation and handling and generate fault isolation and recovery operations:

based on the automatic alarm and processing scheme, adopting a data analysis technology to identify key fault characteristics and generating a fault characteristic data set;

Based on the fault characteristic data set, a support vector machine algorithm is adopted to generate a fault isolation model;

based on the fault isolation model, model training and adjustment are carried out, fault isolation precision is guaranteed, and an adjusted fault isolation model is generated;

based on the adjusted fault isolation model, fault isolation and recovery operations are implemented;

8. The automated power dispatching system-based troubleshooting method of claim 1 wherein, based on the fault isolation and restoration operations, a genetic algorithm is employed to perform resource-optimized dispatching and generate a resource-optimized dispatching plan:

based on the fault isolation and recovery operation, adopting a resource state analysis technology to evaluate resource allocation requirements and generating a resource allocation requirement analysis report;

generating a resource scheduling optimization model by adopting a genetic algorithm based on the resource allocation demand analysis report;

based on the resource scheduling optimization model, model training and optimization are executed, resource allocation efficiency is ensured, and an optimized resource scheduling model is generated;

Based on the optimized resource scheduling model, implementing resource optimization scheduling to generate a resource optimization scheduling plan;

9. The automated power dispatching system-based troubleshooting method of claim 1 wherein policy generation and risk assessment are performed using a reinforcement learning algorithm based on the resource optimized dispatching plan, and protection policies and risk assessment are generated:

based on the resource optimization scheduling plan, evaluating the current resource scheduling environment by adopting an environment analysis technology, and generating a resource scheduling environment analysis report;

simulating a differentiated resource scheduling strategy by adopting a reinforcement learning algorithm based on the resource scheduling environment analysis report to generate a strategy simulation result;

based on the strategy simulation result, carrying out strategy effect evaluation and risk analysis to generate a strategy effect evaluation report;

comprehensively evaluating multiple strategies and selecting an optimal strategy based on the strategy effect evaluation report, and generating a protection strategy and risk evaluation;

10. The power dispatching system-based automated troubleshooting method of claim 1 wherein based on the protection strategy and risk assessment, a simulated annealing algorithm is employed for system optimization and feedback and system optimization and performance feedback is generated:

based on the protection strategy and the risk assessment, analyzing the current system performance by adopting a performance monitoring technology, and generating a system performance analysis report;

based on the system performance analysis report, adopting a simulated annealing algorithm to optimize system configuration and performance and generating a system optimization scheme;

based on the system optimization scheme, executing optimization operation and adjustment, maintaining the optimal performance of an operating system, and generating an execution result of the optimization operation;

based on the execution result of the optimization operation, performing system performance evaluation and feedback to generate system optimization and performance feedback;