On the Applications of Entropy Concept in the Fault Diagnosis and Tolerant Control for Stochastic Systems

Over the past decades,fault diagnosis and fault tolerant control for dynamic stochastic systems have always been one of the important areas of research in control theory and applications.The purpose of fault diagnosis is to obtain effective fault estimation algorithm so that the variance of the residual signals is minimized.Such methods only applicable to Gaussian residuals or the residuals whose probability density functions are of a symetrical distribution shape.However,for non-Gaussian residuals,the embedded uncertainties cannot be fully described by only using their variances.This paper proposes a novel preliminary methodology on fault diagnosis and tolerant control for non-Gaussian dynamic stochastic systems,where the idea is to ensure that the entropy of residual signals for fault diagnosis and the entropy of the closed loop tracking errors are all minimized when the system is subjected to faults.