Research On Fault Diagnosis And Location Method Of Islanded DC Microgrid Line

DC microgrid has the advantages of simple control,low cost and low loss,and has gradually become an important form of power supply and consumption in the future.However,DC microgrids are susceptible to failures due to life expiration,performance degradation,climatic conditions,and natural disasters,which seriously threaten the safe,reliable,and stable operation of the system.The fault characteristics of the DC microgrid are not obvious,the fault time is short,and the overcurrent capacity is low,so the fault diagnosis and location methods of the AC microgrid and the traditional DC system cannot be used.Therefore,it is of great significance to study the fast and accurate fault diagnosis and location methods of DC microgrid.In this paper,the fault diagnosis and location of DC microgrid lines in radial islanded are studied.The main work is summarized as follows:(1)Aiming at the problem of complex line fault mechanism and large characteristic difference of islanded DC microgrid,based on the fault transient modeling and analysis of buck-boost short-circuit fault,the critical time for short-circuit fault diagnosis is given,the rapidity requirement for fault diagnosis of short-circuit fault is revealed.Based on Matlab/Simulink,a simulation model of DC microgrid system based on AVSG control was built.Taking the capacitor voltage and inductor current of each distributed power source as the characteristic quantities,the characteristics of the short-circuit fault and grounding fault of the system under different operating scenarios were analyzed in detail.It provides data and decision basis for fault diagnosis and fault location.(2)In order to solve the problem that short-circuit fault diagnosis requires high speed and high impedance faults are difficult to diagnose,a hierarchical unified fault diagnosis framework based on local information is proposed.Place short-circuit faults and ground faults on the first and second layers of diagnosis,respectively.Aiming at the rapidity of shortcircuit fault diagnosis,a fast diagnosis method based on multiple BPNN is proposed.On the basis of the first-level diagnosis,aiming at the problem that the characteristics of high impedance faults are not obvious and difficult to diagnose,a second-level long-time-scale fault diagnosis method based on fault feature fusion and SVM is proposed.The second layer of diagnosis performs wavelet decomposition and singular value decomposition of fault voltage and current signals respectively,and fuses the extracted relative wavelet energy with the selected singular value to drive SVM to achieve accurate diagnosis of high impedance faults.The effectiveness of the proposed method is verified by simulation.(3)Aiming at the problems of difficult location of high impedance faults,low location accuracy and high sampling frequency under noise pollution,a location method based on wavelet sliding window energy and LSTM neural network is proposed,which improves the performance of faults with no obvious features in noise environment.Aiming at the shortcircuit fault,an short-circuit fault location method based on BPNN is proposed to achieve fast and accurate fault location under limited fault information and noise pollution.The effectiveness of the proposed method is verified by simulation.