Fault Diagnosis for Uncertain Steer-by-Wire System via Uncertain Bond Graph and the Adaptive Difference Evolution Mutation Operation

This paper proposes a fault diagnosis method for steer-by-wire (SBW) system with parameter uncertainty. First, the diagnostic bond graph in linear fraction transformation form is employed to model the uncertain SBW system. The analytical redundancy relations (ARRs) and the adaptive thresholds can be derived to detect the occurrence of fault according to this model. Then, the fault signature matrix can be constructed based on ARRs for fault isolation, and the fault isolation result is the possible fault set (PFS). To refine the PFS and locate the real fault, a new fault estimation method named adaptive differential evolutionary biogeography-based optimization (i.e., ADEBBO) algorithm is proposed where the random mutation operation in the traditional biogeography-based optimization algorithm is replaced by adaptive difference evolution mutation operation. The accuracy of fault estimation is improved by the ADEBBO. Finally, the effective of proposed fault diagnosis method is verified by simulation.

I. Introduction

As a novel trend of the automobile industry development, steer-by-wire (SBW) technology has attracted widely attention due to its better vehicle handing stability, steering performance and driving comfort [1]. SBW system is an electronical controlled vehicle system, which obtains driver’s intent through sensors and transmits control signals to actuators by wire [2]. However, this will increase the failure rate due to the system complexity. Therefore, fault diagnosis for SBW system is essential to ensure driving safety.