The Power System Fault Diagnosis Method Based On Comprehensive Relative Entropy Of S-Transform And Continuous Hidden Markov Model

Nowadays,the scale of the power grid is continuously expanding,and the cross-regional interconnection between power grids is getting closer and closer.When the power grid fails,once it is not processed in time,it may cause a large-scale power outage.When the power grid fails,a large amount of fault information is flooded into the dispatch center.How to quickly identify the fault line based on the fault information and determine the fault type of the power transmission line is the goal of the power grid fault diagnosis method.Therefore,this paper designs two aspects of fault diagnosis about how to identify the fault line and how to identify the fault type of the transmission line which called the fault phase selection.In the past,fault identification methods for power grids were mostly based on switching quantities.The electrical quantity has unparalleled advantages.Therefore,this article uses electrical quantities as fault information to identify faulty lines.The fault phase selection method of transmission line mainly needs to pay attention to two aspects,one is the signal feature extraction method,and the other is the design of the fault phase selection algorithm.Therefore,this paper uses S-transform to extract the characteristics of the signal,and adopts a Continuous Hidden Markov Model with good state recognition effect to design the fault phase selection method.This paper proposes a power grid fault identification method based on comprehensive relative entropy of S-transform and k-means clustering algorithm.According to the fault recording data,the characteristic current data of two cycles before and after the fault moment of the suspected faulty line and the reference line are extracted.Calculate the relative entropy of the S-transform amplitude and the relative entropy of the S-transform energy between the suspected faulty line and the reference line.The comprehensive relative entropy of S-transform is constructed to quantify the difference between the suspected faulty line and the reference line.Then,the k-means clustering algorithm is used to cluster the data of the comprehensive relative entropy of S-transform between the suspected faulty line and the reference line.The results are divided into two types: normal line type and fault line type.After clustering,the data class about the S-transform comprehensive relative entropy of the suspected faulty line with large cluster center value is the fault line.A large number of simulation experiments are performed on the PSCAD IEEE39 node system.The experimental results show that the method is not affected by the change of the fault conditions,and the fault line can still be correctly identified under the conditions of high packet loss rate and noise.This paper presents a fault phase selection method for transmission lines based on S transform and Continuous Hidden Markov Model.Firstly,S-transform is used to extract fault features of transmission lines,and then ten Hidden Markov Model is trained of ten different transmission line fault types and one Hidden Markov Model of normal states according to the extracted fault feature quantities.Each Continuous Hidden Markov Model trains the relevant parameters of the model through modified Baum algorithm.After training,Viterbi algorithm and forward-backward algorithm are used to select the fault of the transmission line.Simulation experiments were conducted on test samples of different fault conditions.The simulation results show that the method has high accuracy of fault phase selection.Even with the addition of 40 dB SNR noise,the method still has a high rate of fault phase selection.