On-board fault diagnosis is critical for the automotive industry. Recently,
new on-board diagnostic system requirements (OBD II) have been enforced on
California vehicles and new legislation will become stricter and stricter,
moreover such requirements have also been extended in Europe (EOBD). Gover
nment regulations will require monitoring vehicle emissions and alerting th
e driver if the exhaust after-treatment system is not working properly. To
meet these requirement, sophisticated diagnostic algorithms have to be deve
loped. This paper presents a model-based stochastic approach for fault dete
ction with application to automotive exhaust-gas after-treatment systems. T
he algorithm, based on relatively simple control-oriented models of the thr
ee-way catalytic converter and the oxygen sensor, is suitable for real-time
, on-board applications. The overall strategy has been tuned and validated
on the basis of experimental data, Copyright (C) 2001 John Wiley & Sons, Lt
d.