This paper presents the application of detection filters to the diagnosis of sensor and actuator failures in automotive control systems.
The detection filter is the embodiment of a model-based failure detection and isolation (FDI) methodology, which utilizes analytical redundancy within a dynamical system (e.g., engine/controller) to isolate the cause and location of abnormal behavior (i.e., failures).
The FDI methodology has been used, among other applications, in the aerospace industry for fault diagnosis of inertial navigation systems and flight controllers. This paper presents the philosophy and essential features of FDI theory, and describes the practical application of the method to the diagnosis of faults in the throttle position sensor in an electronically controlled IC engine.
The paper also discusses the incorporation of FDI systems in the design process of a control strategy, with the aim of increasing reliability by embedding diagnostic features within the control strategy.