Satellite Fault Diagnosis Method Based On Evidence Theory

The application of satellite system has penetrated into various fields such as satellite communication,logistics engineering,measurement and monitoring.Fault diagnosis is an important research field of satellite system.In order to guarantee the availability and the quality of service,in the event of failure,the satellite system should be able to independently carry out fault diagnosis,which helps locate and repair the fault accurately.When the system has not experienced a significant failure,the system should be able to independently assess the performance of the current system of health,what nip problems in the bud.This thesis focuses on technologies of fault diagnosis for satellite system.In the fault diagnosis of fault classification,a fault diagnosis method based on improved D-S evidence theory is proposed.Firstly,the method determines the typical cases of each fault type and calculates initial probabilities of failure according to the parameters in the case.Then the fusion probability of each parameter is obtained to get the final diagnosis probability.Finally,comparing the probability value of each case to determine the type of fault diagnosis.When determining the parameter weights,the method adaptively allocates the parameter weights according to the parameter distance.The simulation results show that the proposed approach has a high diagnostic accuracy and reliability.In the field of performance evaluation,this paper presents a fault diagnosis method based on evidential reasoning(ER).Firstly,the rating is divided into three performance grade.Secondly,the performance parameters of the satellite engine are transformed into the reliability division.Thirdly,by using the analytic ER algorithm to fuse initial reliability,the overall reliability of each performance grade was diagnosed.Finally,diagnosis results of reliability distribution are quantized using utility theory.This utility represents overall preference degree which can determine whether the system failure.When the parameter weight is determined,the particle swarm optimization algorithm is used to optimize the weights.The simulation results show that this method is more accurate than single parameter diagnosis and fuzzy information extraction.