Research Of Spacecraft Survivability Acessment Expert Systems Based On Fuzzy Bayesian Network

Space debris is seriously threatening the orbital operation of the spacecraft.The average velocity of space debris hitting the spacecraft exceeds 10km/s.Such an ultrahigh-speed impact process can easily penetrate the spacecraft protection structure once it occurs.Lightly,the spacecraft is partially ineffective and functionally reduced,and the single-machine disintegration machine is destroyed.Spacecraft survivability assessment in space debris environment is of great practical significance and practical value for mission risk assessment and mission planning.Analyze and compare the original spacecraft survivability assessment and fault diagnosis methods,and clarify the problems existing in the original method,and propose a set of expert systems for evaluating spacecraft survivability,using Bayesian networks,fuzzy Bayesian network algorithm for reasoning.In view of the fact that the original evaluation model does not have completeness,practicability and accuracy,this paper develops a spacecraft survivability assessment expert system.A friendly and friendly human-computer interaction interface is designed for knowledge system dialogue.The knowledge acquisition machine specification knowledge acquisition is designed based on the hierarchical model of fault tree.The spacecraft component failure mode is used as the bottom event,the subsystem failure is used as the intermediate event,and the system failure is taken as the system failure.The top event completes the description of the system failure risk transfer process;the knowledge is stored in the library,the knowledge expression method of production rules is used,and the knowledge storage process is completed in combination with the database;the knowledge management function is realized based on the maintenance of the fault tree model.In order to solve the problems of complex calculation,one-way reasoning and poor accuracy of the original fault tree analysis method,Bayesian network inference method is used to complete expert reasoning.The directed acyclic graph structure is constructed by the fault tree mapping,the confidence degree of the production rule is used as the conditional probability of the directed edge,and the failed component and the failure probability are used as the activation inference rule.By using the two-way probabilistic reasoning method of Bayesian network,the full probability inference results of subsystem,system failure probability and system survivability are obtained,and the weak link of system survivability is determined by solving the posterior probability.Due to the "two-state" of the viability assessment and the uncertainty of the conditional probability of the risk transfer model,this paper proposes a viability assessment method based on fuzzy Bayesian network inference.The fuzzy probability is mainly processed for the expert experience description of Bayesian network node parameters,and the Bayesian network two-way reasoning is also carried out,which improves the accuracy of the viability evaluation result.Combined with the spacecraft FMECA,the rating of system vulnerability and the evaluation of spacecraft survival index are proposed to realize the transition of survivability from conceptualization to practical application.Software development for the above design scheme,using Visual Studio 6.0 as the operation interface for human-computer interaction,data interaction,expert system development platform,background database and management selected SQL Server 2014,BN,FBN algorithm reasoning in Matlab 2016 a Development calculations.The analysis results are consistent with the results of the classical survivability algorithm,which proves the correctness of the survivability algorithm proposed in this paper.The method provides an efficient spacecraft survivability evaluation scheme,which is not only compatible with the risk assessment under the impact of space debris,but also compatible with the evaluation of various damage mechanisms,and has certain engineering application value.