Detection and isolation of faults in the exhaust path of turbocharged automotive engines

Detection and isolation of faults in the exhaust gas path of a turbocharged spark ignition (SI) engine is an essential part of the engine control unit (ECU) strategies to minimize exhaust emission and ensure safe operation of a turbocharger. This paper proposes a novel model-based strategy to detect and isolate an exhaust manifold leakage and a stuckclosed wastegate fault. The strategy is based on a globally optimal parameter estimation algorithm which detects a virtual hole area in the exhaust manifold. The estimation algorithm requires observation of the exhaust manifold’s input and output flows. The input flow is estimated by a nonlinear Luenberger observer which is analytically shown to be robust to the faults in the exhaust manifold. The output flow of the exhaust manifold is estimated by a sliding mode observer. The designed fault detection and isolation (FDI) strategy is tested with the experimental data collected from a 1.7-liter turbocharged SI engine. The validation results show that the FDI strategy can successfully detect a leakage fault caused by a 5 mm hole in the exhaust manifold, and can identify the wastegate stuck-closed fault.