Model-based fault detection and isolation for electric power steering system

Abstract

Electric power steering (EPS) system is an advanced steering system that assists steering torque using the electric motor. To guarantee reliability and safety of EPS System, fault management is especially important. This paper provides a fault detection and isolation (FDI) technique for the sensors of EPS system through the model-based FDI method. In the presented work, FDI algorithm is composed of a two-step process by using the mathematical models of the electric motor and the steering system. In the first step, faults of the phase current sensor and the angular velocity sensor of the electric motor can be detected by using parity equation which is derived from the mathematical model of the electric motor. In the subsequent step, residual is produced by cross-checking with three torque signals in mathematical model of the steering system, estimated assist torque from the measured phase motor current, estimated road torque from the tire model, and driver's wheel steering torque. And then the steering sensor fault can be detected by comparing the residual with the threshold which includes model uncertainties. Through the both steps, the EPS system sensor fault can be isolated by identifying the faulty component.