A Stochastic Modeling Approach of Quantized Systems with Application to Fault Detection and Isolation of an Automotive Electrical Power Generation Storage System

Abstract

This paper introduces a stochastic modeling approach for a quantized system for the purpose of fault detection and isolation in an automotive alternator system. Three common alternator faults including belt slip, diode failure, and incorrect reference voltage for the voltage controller are considered and analyzed. A continuous nonlinear model of the alternator system is quantized into discrete states in order to simplify diagnostic efforts. The paper describes a stochastic modeling approach that creates a time-varying probability transition matrix that can be computed in real- time without the need for Monte Carlo simulation. Fault detection and isolation occurs through comparison of the most probable state from the transition matrix and the quantized output state.