Model based diagnosis for automotive engines-algorithm development and testing on a production vehicle

A complete diagnostic algorithm is presented for detecting and isolating faults of actuators and sensors of an automotive engine. The algorithm is built around the concept of analytical redundancy and utilizes structured parity equations. The parity equations rely on a nonlinear and dynamic model, obtained by identification using data from a single production vehicle with a 3.1 l V-6 engine. Tests performed with the diagnostic algorithm on the same vehicle show excellent performance under different fault conditions. Several important issues require further investigation, including long-term climatic effects, vehicle aging, car-to-car variations, the automation of the calibration procedure and implementation in the face of processor limitations