Model-Based Failure Detection and Isolation of Automotive Powertrain Systems

Failure detection and isolation issues of engine speed, torque converter turbine speed, and wheel speed sensors commonly used in automotive powertrain controllers are presented in this paper. Due to the highly nonlinear nature of powertrain dynamics, constant gain extended Kalman filters are designed for use as detection filters. Then, a failure detection and isolation algorithm utilizing functionally redundant sensors is developed and applied to the powertrain problem. The developed algorithm is inherently robust to unmodeled dynamics and process noise and is effective in detecting and isolating failures of the powertrain sensors.