Research On The Fault Diagnosis Technology Of Distribution Network Based On The Combination Of Improved RS-SVM

As an important part of the power system,the distribution network directly faces users,and is a key link to ensure the quality of power supply services and user services and improve the efficiency of power system operations.In the process of rapid development of distribution network automation,ensuring the safety,stable operation,and power supply reliability of the distribution network has always been the top priority of power grid construction and operation.On the one hand,the distribution network automation system provides a wealth of data information for fault diagnosis of the distribution network,which contains rich potential value;on the other hand,the increase and use of a large number of intelligent equipment in the distribution network has led to the distribution network data Information acquisition systems such as Supervisory Control And Data Acquisition(SCADA)collect a large amount of useless redundant fault information,and incomplete fault signals are mixed in,which troubles the fault diagnosis of the distribution network.The complex structure of the distribution network also increases the frequency of multiple faults,making diagnosis more difficult.Therefore,a fast and accurate fault diagnosis method for the distribution network is of great significance to the safe,stable and economic operation of the distribution network.Aiming at the characteristics of incomplete fault information and redundancy of fault information in distribution network,this paper proposes a fault diagnosis method based on improved rough set theory(RS)and support vector machine(SVM).Use the action information of each circuit breaker and protector in the fault information as the input of the fault diagnosis model to locate and diagnose the line or area where the fault is located.The main contents of this article are as follows:First,the use of rough set theory for fault information preprocessing.Fully considering the characteristics of redundant and incomplete fault information,based on the binary discernibility matrix reduction algorithm,an improved attribute reduction algorithm based on generalized binary discernibility matrix is proposed.Establish the original decision table according to the distribution network topology structure relationship,use the improved reduction algorithm to perform attribute reduction on the original decision table,obtain the reduced minimum decision table and decision rules,and ensure the data validity and Under the premise of diagnosis accuracy,the amount of fault information data is reduced.Combined with a simple distribution network example for verification analysis,it laid the foundation for the next fault diagnosis.Then,build an improved RS-SVM combined distribution network fault diagnosis model.In the process of constructing the support vector machine classification model,the particle swarm optimization(Particle Swarm Optimization,PSO)based on two strategies of non-linear inertia weight reduction and asynchronous linear change learning factor is used to optimize the support vector machine kernel function parameters g and Penalize parameter C to improve the shortcomings of traditional support vector machines that are prone to fall into loop optimization.The obtained optimal solution is used for training and learning of the support vector machine,and the reduced test sample decision data is used for fault diagnosis.Finally,the actual 10 kV distribution three-port ring network in a certain area is taken as an example for experimental analysis.According to the single failure and multiple failures,it is compared with the single SVM and RS-SVM algorithms,and combined with a variety of evaluation indicators to verify the feasibility and accuracy of the method used in the distribution network fault diagnosis.