Research On Fault Diagnosis Of Marine Power System

As an important part of the ship,the ship’s power system is responsible for the task of transmitting electrical energy for ship equipment.In the ship’s DC power system,the alternator outputs DC power to the DC power grid through a rectifier,and then distributes it to various electrical equipment.Among them,generators and rectifiers play an important role.If they fail,it will cause the ship’s DC power system to fail,and even seriously threaten the safe operation of the ship.This article focuses on the synchronous alternator and rectifier containing uncontrollable rectifier circuit in the ship’s DC power system.The main research content is the fault diagnosis of synchronous generator and rectifier with uncontrollable rectifier circuit.It uses neural network and optimization algorithm The classification and diagnosis of generator and rectifier faults are as follows:First,based on the relevant parameters,Ansys Maxwell software is used to establish the finite element model of the synchronous generator;the finite element model is imported into the Ansys Simplorer software to build a joint simulation model of the ship’s DC power system;the electromagnetic model and three-phase voltage,The current verifies its accuracy.Secondly,research the generator and rectifier faults,analyze the causes of the faults and the changes in the electromagnetic torque;modify the joint simulation model,complete the simulation tests under different fault conditions,collect the simulation data for analysis and research,and summarize the change law.Then,use Fourier transform to process the collected simulation data,observe its spectral characteristics,and compare with the theoretical analysis;use wavelet packet decomposition to decompose and reconstruct the data,use frequency band energy analysis technology to extract fault feature quantities;use PCA method The feature quantity is processed,and the main component combination is extracted to form the fault feature vector,so as to achieve the purpose of reducing the dimension.Finally,BP neural network and RBF neural network are used to classify and diagnose the fault data,and compare their training performance and diagnosis accuracy rate.For the problem of slow convergence of BP neural network and easy to fall into local minimum,use genetic algorithm to improve The genetic algorithm and particle swarm algorithm are used to optimize the parameters of the BP neural network,and the training performance and diagnostic accuracy of different optimization algorithms are compared;the diagnostic performance of the above five fault diagnosis methods is compared.