Misfire Diagnosis of Diesel Engine based on Rough Set and Neural Network

Abstract Misfire of diesel engine was simulated by the bench test. Top center signal and cylinder vibration signal were collected under different states. The vibration signal from each cylinder's combustion period was intercepted according to the diesel combustion sequence, effective value was calculated. The short-time effective values were reduced by rough sets, and BP Neural Network model built with the reduced character was used to diagnosis diesel misfire. The result showed this method could locate the misfire cylinder correctly, and it was meaningful for guiding the detection and repair of vehicle.