Fault diagnosis of an active suspension control system

Abstract

In this chapter a model-based fault detection applied on a vehicle control system is presented, which relies on simplemathematical descriptions of the system and which yields a robust fault detection and isolation of faults against disturbances or model uncertainties affecting the system. Using multiple models and observer as virtual sensors, which are based on simplified descriptions of the system, symptoms appropriate for fault diagnosis were generated. A fuzzy filter together with a knowledge base describing the relation between the faults and the symptoms were used to detect and isolate the faults as fast as possible while at the same time avoiding false alarms. The goals of the fault diagnosis described in the chapter were achieved by detecting faults that were classified as hazardous, achieving fast and robust fault detection, and by using simple models that enhance the feasibility of the approach for the automotive application. Using explicitly a few pressure sensors for the generation of the residuals, the cost and packaging effort of the diagnosis system is held to a minimum.