A research on the misfire diagnosis method of gasoline engine based on BP neural network

作者：Hu Jie^*; Yan Fuwu

来源：Qiche Gongcheng/automotive Engineering, 2011, 33(2): 101-105.

• BP neural networks
• Characteristic parameter
• Electronically-controlled gasoline engine
• Fault diagnosis
• Misfire

In view of the defect of general misfire diagnosis algorithm based on instantaneous crankshaft angular acceleration that it can diagnose misfire, but can't effectively identify misfire fault mode, a new misfire diagnosis algorithm is proposed based on the time of work and BP neural network. According to the fluctuation of the time of work in each cylinder under different modes, the characteristic parameters of the time of work signal in each cylinder in a diagnostic cycle are extracted; and combining with the identifying function of BP neural network, misfire diagnosis in different fault modes is realized. Bench tests are carried out to detect the misfire diagnosis situation at four different modes, i.e. normal operation, single misfire in 3rd cylinder, successive misfire in both 2nd and 3rd cylinders and successive misfire in 3rd cylinder. The results shows that the diagnosis algorithm presented can identify different misfire fault modes and locate the misfired cylinder effectively.

• 出版日期2011