Analysis of model-based sensor fault diagnosis with application to a motor-driven power steering system

Abstract

This paper presents a theoretical analysis of model-based fault detection and isolation (FDI) and its application to a motor-driven power steering (MDPS) system. In particular, this paper proposes a systematic approach to determining fault detectability and isolability of a linear time-invariant system based on a parity equation. The proposed approach is applied to the MDPS system in order to examine the fault detectability and isolability of the MDPS depending on the number and type of sensors and disturbances. Fault detection and isolation of the phase current sensors and the angular velocity sensor are conducted using a parity equation derived from the mathematical model of the permanent magnet alternating current (PMAC) motor. The fault of the steering torque sensor is detected and isolated by cross-checking with two different torque estimates. An adaptive threshold is proposed to improve the performance and robustness of the proposed FDI method against time-varying model uncertainties. Simulation and vehicle test results show the validity and performance of the proposed FDI method for the MDPS system.