Adaptive Threshold Based Diagnostics for Steer-By-Wire Systems

Abstract

In the process of implementing modular fault diagnostics for X-by-Wire systems (Pisu et al., 2000), it was discovered that distinguishing between the model uncertainties and occurrence of faults is a real challenge. It is imperative to solve the problems that exist in using fixed observer and threshold, including the observer model mismatch, generated threshold misfit, diagnostic parameter uncertainties, and incorrect diagnostic output during high stress but normal vehicle operation. Compared to the fixed observer and threshold strategy, the improved adaptive threshold-based diagnostics described in this report are more robust and applicable because of the addition of adaptation into the diagnostic observer and threshold generator. The report also describes an approach to fault detection and isolation in the presence of model and parameter uncertainties. This approach has been successfully implemented in the NAVDyn (Non-Linear Analysis of Vehicle Dynamics) simulation model using Matlab/Simulink, and simulation results are provided to verify that the strategy and implementation are viable.