Fault Diagnosis Method of Ship Fuel System Based on Kernel Principal Component Analysis and Particle Swarm Optimization Support Vector Machine

Aiming at the strong non-linearity and noise characteristics of fault sample data of marine fuel system, traditional support vector machine (SVM) has low classification accuracy in fault diagnosis of marine fuel system. This paper studies the extraction method of fault sample features and the optimization method of SVM parameters, and proposes a new intelligent fault diagnosis method based on kernel principal component analysis (KPCA) and particle swarm optimization (PSO) to optimize SVM. Firstly, KPCA is used to extract the nonlinear features between the sample data, reduce the data complexity, extract the high-dimensional information of the sample data, and extract the non-linear principal component as the input sample of SVM. Secondly, PSO optimization algorithm is used to optimize the penalty factor and the kernel function parameters of SVM. Finally, the optimized SVM model is used to identify the reconstructed samples. The results of comparative experiments on fault diagnosis of ship fuel system show that the method has higher diagnostic accuracy than single SVM fault diagnosis method, which verifies the superiority and validity of the fault diagnosis method.