Research And Implementation Of Ground Fault Diagnosis System Of Flight Control System Based On Symptom Analysis

The new fighter is one of the outcomes for the continuous development of modern military technology.Because of its great strategic deterrent and high equipment value,the requirement of the equipment safety is high.Among the factors which affecting the fighter’s safety,the fighter’s fault is a key factor.Because flight control system is an vital part of the fighter,So it’s fault diagnosis is necessary to study for improving the efficiency of equipment fault diagnosis.The expert and maintenance personnel usually use maintenance built in test system to diagnose the fault in the flight control system on ground.Expert and maintenance personnel can use it to analyze the symptom of fault and diagnose the fault.However,the manual symptom analysis and the requirements of professional knowledge lead to a decline in the efficiency of fault diagnosis on ground.The modern intelligent fault diagnosis method is studied and used in this thesis to built a ground fault diagnosis system which has the ability to locate the fault by reading and analyzing the symptom automatically,thus reduce the human participation and improve the fault diagnosis efficiency.This thesis mainly includes the following several aspects:Firstly,the research status of the current fault diagnosis method is studied.And according to the compositional structure of the flight control system,the fault characteristics of the flight control system and the mapping relations between the fault and the fault symptom are analyzed.Then the requirement of the intelligent maintenance built in test system of flight control system based on symptom is analyzed.Secondly,through the research on the methods used for fault diagnosis,it has been found that after a long period of using,there is a lot of knowledge of fault diagnosis such as documents and experiential rules which describes the mapping relations between the fault and the fault symptom.Based on the form of these knowledge,the corresponding knowledge base is designed,and a rule learning machine which can learn the knowledge by rules’ graphical modeling and loading document descriptions is designed.Then the system’s reasoning machine which is efficient for the knowledge base in this thesis is designed with the hybrid reasoning method.And the system’s explaining mechanism which is formed by text-prefabricating and path-tracking method is designed.All of these form a rule-based method for flight control system’s fault diagnosis which can learn and use these knowledge efficiently.Thirdly,based on the comparison and analysis of error back propagation neural network,radial basis function(RBF)neural network and support vector machine,the RBF network is introduced to learn the mapping relations between the fault and the fault symptom which the method based on rule can not describle,thus the fault can be diagnosed by symptom analysis.Then the training method of RBF network is studied.A modified hierarchical genetic algorithm is proposed to train the RBF network.Hierarchical genetic algorithm is improved by initializing the population based on the training samples and using the differential evolution operator in the crossover operation.The evolutionary algebra of the population is reduced by using high fitness initial population which is initialized by training samples.The diversity of the population is enhanced by the differential evolution operator.Thus premature problem was solved because of the enhanced diversity.The performance of training algorithm based on hierarchical genetic algorithm is improved by these inprovements,thereby the fault diagnosis performance of RBF network is improved.Fourthly,the general scheme of fault diagnosis which combines the rule-based fault diagnosis method and radial basis function(RBF)neural network fault diagnosis method to diagnose the fault is proposed.The complete software design and implementation of the ground fault diagnosis system of flight control system based on symptom analysis is proposed,and the effectiveness of the system is verified on the aircraft simulation platform.