RESEARCH ON FAULT DIAGNOSIS OF ELECTRICAPPLIANCE FOR VEHICLE BASED ON CAN BUS

摘要

Aimed at the problem and difficulty of diagnosis after thernevent in the traditional form of the fault diagnosis of electricrnappliance for vehicle , a new fault diagnosis system isrnpresented which is based on the combining of CAN bus andrnembedded database. Using a probabilistic neural networksrn(PNN) algorithm, the state parameters of engine were obtainedrnby CAN bus and were used to identify multiclass-state withrnthe samples prestored in the embedded knowledge base, andrnwith the exceptional signals from the sensors, a contingentrnfault can be estimated and determined effectively. Thernalgorithm has shorter training time and higher right diagnosticrnlevel compared to Back-Propagation neural networks. Arnmethod of the fault diagnosis of self-determination andrnintellectual for vehicle is realized, as the result, the handlingrnperformance of vehicle is improved, and the probability ofrnfault present can be decreased, so does the maintenance cost.