Research On Fault Diagnosis Of Three-phase Inverter Based On Particle Swarm Optimizes Neural Network

With the development of power electronics technology,Voltage Source Inverter(VSI)has been widely used in household appliances,new energy vehicles,rail transit and electrical transmission systems,etc.As the core device of power conversion in power equipment,the reliability and safety of inverters are very important.Therefore,it has important research value for inverter fault diagnosis methods.This paper mainly studies the three-phase VSI open-circuit fault diagnosis method based on neural network.Aiming at the defects of traditional neural network,a neural network fault diagnosis method improved based on different optimization algorithms is proposed,and the test results are compared and analyzed.The main research contents of this paper are as follows:The various faults and hazards of three-phase VSI are introduced,and most research value of the common IGBT open circuit faults were classified.Using MATLAB/simulink to simulate the open circuit fault,and analyzing the characteristics of the output voltage after the fault.Considering the influence of noise on the output voltage and increasing the difficulty of fault diagnosis,this paper adds Gaussian white noise to the output voltage as the original fault signal.Using Fast Fourier Transform to extract characteristic information from the original fault signal: DC component,fundamental wave amplitude,second harmonic and third harmonic amplitude and phase angle,and then normalize the fault information.The wavelet packet analysis method is used to decompose and reconstruct the original fault signal,and the energy distribution value of each frequency band is extracted as the fault characteristic information.A BP neural network model suitable for three-phase VSI open-circuit fault diagnosis was built,and the neural network was trained using the fault information obtained through feature extraction.According to the change curve of cross entropy loss value,the ability of different fault information to reflect different faults is compared and analyzed.Considering that the particle swarm optimization(PSO)algorithm has a powerful global search capability,it is just suitable for combining with the local search capability of the BP neural network.Therefore,the PSO algorithm is used to optimize the BP neural network,and the training effect of the network before and after the optimization is analyzed.The test results show that Using PSO algorithm to optimize BP neural network can improve the training speed and training effect of the network.The derivation formula and basic process of the particle swarm optimization algorithm(BAS-PSO)that incorporates the search behavior of longhorn beard are described.The BAS-PSO algorithm is used to optimize the BP neural network,and the fault diagnosis effect of the BP neural network based on the BAS-PSO algorithm and the PSO algorithm is compared and analyzed.The test results show that the three-phase VSI open-circuit fault diagnosis method based on the BAS-PSO algorithm to optimize the BP neural network has a higher diagnosis rate and robustness.Aiming at the problems of the neural network-based fault diagnosis method,the correct rate of Top K is proposed to solve the problem,and the feasibility of the method is verified by experiments.