DIESEL ENGINE FAULT DIAGNOSIS BASED ON DT-CWPT AND RBF NEURAL NETWORK

Abstract

Aiming at the large amount of vibration signal and data redundancy on the cylinder head of diesel engine, this paper uses DT-CWPT to process the acquired signal, including data denoising processing and feature vector extraction. After the wavelet decomposition is collected, the dimension of the signal is reduced, and the excess signal components can be filtered out, the fault features are highlighted, and the information contained in the signal is not damaged, and the accuracy of the fault diagnosis is improved; the RBF neural network has an excellent mode. Recognition performance, relative to the neural network has a rapid diagnosis ability; particle swarm optimization algorithm to optimize the RBF neural network basis function, can improve the diagnostic speed of RBF neural network. Finally, the research is applied to the actual experiment to verify the superiority of the method.