The Active Fault Diagnosis Method Researches Based On Auxiliary Input Signal

Fault detection and diagnosis play an important role of improving the safety,reliability and maintainability of the operation for automatic equipments.Generally speaking,the minor faults are the precursors of significant faults.The more timely of the fault,the more accurate of the fault detection result,the more helpful to the safe and stable operation of the system.In the existing methods,the passive approaches are generally adopted,however,it is difficult to detect minor faults exactly and punctually.Thus,this paper adopts active fault diagnosis mechanism and designs the on-line auxiliary input signal for the purpose of detecting system faults.This paper puts forward two design strategy of designing auxiliary input signal for minor fault detection and isolation problem.Firstly,this paper presents an on-line auxiliary input signal design strategy based on ellipsoid idea in set-membership approach and moving window techniques to deal with the problem of active fault detection and isolation.The output sets are characterized by ellipsoids based on set-membership and moving window techniques,further,active fault detection and isolation problem is established.Due to the constraints for the intersections between different sets are empty are hard to solve,then the condition is converted to the inequality form to solve the problem.Further,this problem is solved by a two-step strategy,that is the auxiliary input signals are designed on-line by considering active fault detection condition,and it is isolated by judging the actual system output belongs to which output ellipsoid of the faulty models,or determining the probability of the system output is in which faulty model when the output ellipsoids of faulty models are intersecting.Finally,in this paper,an on-line auxiliary input signal design strategy for discrete-time system is proposed based on zonotope and moving window techniques to solve the problem of on-line active fault detection and isolation.The output sets are described by zonotopes and the optimization problem is established.To facilitate the auxiliary input signals design,the optimization problem is converted to the MIQP problem.Further,the system output belongs to which system model can be determined by the fault detection and isolation logic.It is compared with the ellipsoid-based auxiliary input design approach,the simulation experiment is adopted to demonstrate the effectiveness of the zonotope method.