Model-based engine fault detection and isolation

Abstract

To a large extent, tailpipe emissions are influenced by the accuracy and reliability of the intake manifold sensors and the predictive models used for cylinder charge estimation. In this paper, mathematical models of an internal combustion engine are employed to detect failures in the intake manifold. These can be associated with the upstream sensors such as the pressure and temperature sensors as well as systemic faults such as a leakage in the intake manifold. Any fault will adversely affect the proper operation of the air-fuel ratio control system and must be detected at an early stage. Through the use of dedicated observers, residual errors can be generated and thresholds established. Methods for the isolation of the detected faults are proposed and applied to a 5.7 L V8 engine model. Simulation results for the Federal Test Procedure (FTP) driving cycle indicate that fast and reliable detection and isolation of the faults is possible.