Model-Based Fault Detection and Isolation in Automotive Electrical Systems

Abstract

This paper deals with the design and validation, through simulation in a Matlab/Simulink and SABER environment, of model-based diagnostic algorithms for an automotive electric power generation system (EPGS). The EPGS includes alternator with rectifier, a battery, and a voltage regulator. The mathematic models of these subsystems, based on the physics of the processes involved, consist of time-varying nonlinear ODEs. The diagnostic problem focuses on the detection and isolation of a specific set of alternator faults, including belt slipping, rectifier fault and voltage regulator fault. An in-depth analysis of the models is conducted in order to understand the effects of different failure modes on system performance; subsequently, an equivalent input-output model of the alternator is formulated and parameterized. The equivalent model permits considerable simplification of the algorithms. The proposed diagnostic approach is based on the generation of residuals obtained using system models and comparing the predicted and measured value of selected variables, including alternator output current, field voltage, battery current, battery voltage and battery temperature. This paper presents the models, diagnostic algorithms and simulation results.