A Mathematical Model Based Method for Diagnosing Failures in Automotive Electronic Systems

With the increasing use of complex electronics in automobiles, diagnosis of failures (including calibration shift) of components has become a major issue. This paper presents a method which has the capability of diagnosing failures of any component at any level of failure for automotive electronic systems.

The method presented here is based upon mathematical models for the system under diagnosis. These models are run onboard the vehicle in a digital computer and implemented with such highly efficient algorithms that real time diagnosis is possible onboard the vehicle. Although the details of the method are specific for failures in each major class of component (e.g. sensor, actuator, controller), this paper demonstrates that any failure which can be represented in the model can be diagnosed uniquely and distinguished from other failures.