The Modeling Procedures for Model-Based Diagnosis of Slowly Changing Fault in Hybrid System

Abstract

Many traditional model-based diagnosis is based on the assumption of abrupt faults. This assumption can lead to a wrong result when the faults are slowly changing. The diagnosis of slowly changing faults can be considered to be an open area in controlling complex system of large scale. The slowly changing faults are difficult to detect, extremely when interlacing with abrupt faults. The modeling of diagnosis in slowly changing faults is different with abrupt faults. A modeling method of slowly changing faults on hybrid system is proposed, based on Hybrid Input Output Automata and mixture bond graph. The effectiveness is proved by method testing on three-tank system. The testing results provide a basic model to faults discovering and isolating