On-Line Sensor Diagnosis of the Diesel Engine Cold Starting Based on RBFNN

Based on the radial basal function neural network (RBFNN) and the OLS arithmetic, an on-line sensor fault detection diagnostic strategy on the diesel engine cold starting is proposed. This diagnostic strategy is conducted by RBFNN .The data of sensor sampling is the input and the sensor faults is the output of RBFNN. Some samples could be trained and studied by RBFNN. The parameters, such as, short circuit, open circuit and the stuck-at fault of the electric current, the voltage and the rotational speed have been made by the RBFNN and the OLS arithmetic. The test results indicate that the sensor fault diagnostic accuracy can reach 95.6%. It is value that this diagnostic strategy could be achieved the IV emissions regulations of China in the diesel engine cold starting cycle and also can be used in the vehicle on board diagnosis system.