关键词: 故障诊断, CAN总线, 主成分分析, 减法聚类, 自适应神经模糊推理系统

Abstract: Based on CAN bus, a new online PSA (PCA-subtractive-ANFIS) fault diagnosis model is proposed to solve the problems that the conventional engine fault diagnosis cannot be predicted and are hard to inquire about their happenings. With the state data acquired by ECU (electronic control utilities) sensors and transmitted by CAN bus for all the elements of an engine taken as samples for diagnosis, the diagnosis model is developed the way the PCA (principal component analysis) is used for dimensions' reduction and decorrelation and the subtractive clustering algorithm is introduced to complete the fuzzy inference process, then the adaptive-network-based fuzzy inference system (ANFIS) is applied to the new type model. Simulation results showed that the PSA model is superior to PCA-BP network model in fitting, denoising and convergence rate.