Evaluation of Observers for Fault Diagnosis on an Automotive Engine
Fault diagnosis is an important area in automotive applications. A
diagnosis system is taken forth using the method of structured
hypothesis tests. The test quantities used in the hypothesis tests are
calculated from prediction error from fault models. A parameter in
each model is estimated using a state space observer, where the
parameter is viewed upon as an extra state with zero time
derivative. Two observer design methods are compared: linearisation of
the system in combination with pole placement and an observer with
feedback gain only on the fault parameter state. Simulations based on
real data from an automotive diesel engine indicates that the method
of linearisation in combination with pole placement gives the better
diagnosis system of the two tested approaches. However, the method of
feedback gain only on the fault parameter is less complex and easier
to implement.
Niten Olofsson
2000
Informationsansvarig: webmaster
Senast uppdaterad: 2021-11-10