Research on Fault Diagnosis Method for Diesel Engine Fuel System based on Model-Softmax

In order to improve the fault diagnosis accuracy of diesel fuel system and save resources consumed by the fault diagnosis algorithm in the embedded system, a two-stage fault diagnosis method based on engine's dynamic model and Softmax regression algorithm is proposed in this article. First, a model-based active disturbance rejection control algorithm is designed according to the established simplified model of fuel system from diesel engine. Furthermore, the fuel leakage observer is constructed to identify the occurrence of faults by observing the variation of fuel leakage. In the end, Finally, the physical model parameters of the fuel system are extracted by the recursive least square method as fault feature parameters, which are send to Softmax classifier to diagnose fault states of the fuel system. Five categories of faults under different working conditions are verified on the diesel fuel system simulation platform. Verification results show that the accuracy of fault diagnosis from fault characteristic parameters extracted by physical model is 5.9% higher than that from rail pressure signal. Moreover, two-stage fault diagnosis method reduce computing resources consumed in embedded systems.