Application of sliding mode observers to automobile powertrain diagnostics

This paper presents a model based approach to the problem of automobile powertrain monitoring. A model discrete crank-angle domain model is derived. A procedure for designing discrete sliding mode observers, for state and input estimation is presented and applied to the powertrain estimation problem. The sliding observers are used in the nonlinear parity equation residual generation scheme for fault detection and isolation. Experimental results are provided to demonstrate that the designed observers estimate the powertrain states with good accuracy and that the monitoring scheme is effective in isolating both actuator and sensor faults