Electronically Controlled Gasoline Engine Sensor Fault Diagnosis Based On Bp Neural Network And Regulation

As the emission control regulations become more and more stricter, the electric control gasoline engine has become the standard equipment of light automobile. In 2008 year, the Third Country Standard Regualition for Vehicle Pollutant Discharger will come true and On-Board Diagnosis (OBD) systems are forced to set. In the last decade an increasing number of sensors and actuators have been employed in automotive systems. The first reason is that the correct operation of modern electronically controlled engines is based on a high number of measurements. Moreover the growing demand on security and comfort pushes toward an increasing use of suitable sensors and actuators. So many scholars pay maore attention to the on-board fault diagnosis for engine sensors.The paper deals with the hybrid solution based on Value-check, artificial Neural Networks(ANNs) and inference rules adopted in the realization of three sensors(crankshaft position sensor(CPS), throttle position sensor(TPS), inlet manifold air pressure(MAP)fault detection, isolation and accommodation scheme for electronic control gasoline engine applications. In this scheme, short circuit, open circuit faults are detected based value-checked; residuals are generated as difference between measured sensor outputs and corresponding expected value provided by suitable fault detection ANNs, Thus sensor uncalibration, value holding faults are detected. Once fault on TPS or MAP was detected, the fault accommodation ANN, which corresponds to the detected faulty sensors, is started at time. Through bench experiments sampling data, training & testing the ANNs are given together with a good performance of this scheme in this paper.