Research On Bayesian Network For Satellite Uncertainty Design Optimization

As the overall performance of satellite system becomes higher and higher,strict requirements are put forward for the overall design of satellite system.In the overall design stage of satellites,how to analyze and quantify the impact of uncertainty factors on the reliability of satellites and guide the overall design optimization of satellite systems are an important issue to be solved urgently.Bayesian Network is one of the effective solutions to the problem of reliability modeling and analysis of complex satellite systems.However,with the increase of satellite system components,the state composition of components increases exponentially,and even a combination explosion occurs,which leads to great challenges in the modeling and analysis of Bayesian Networks.To solve this problem,an efficient and accurate reliability modeling and analysis method for complex systems is developed based on Bayesian Network and compression reasoning.The main research work includes the following four aspects:First of all,this paper studies the development status and application of Bayesian Network and summarizes the current modeling methods of bayesian network,the application status of bayesian network at home and abroad in the engineering field,the system reliability analysis method based on Bayesian Network,and the advantages and disadvantages of Bayesian Network.Based on the research status and application of Bayesian Network,this paper introduces one of the compression inference algorithms which can solve the combined explosion problem of binary system.Secondly,considering the shortcomings of compression algorithm and inference algorithm of binary system at present,this paper develops an improved compression algorithm which implements the effective compression of arbitrary binary node probability tables.For independence and dependence binary node,an inference algorithm is improved based on improved compression algorithm.On the basis of improved methods,an efficient reliability analysis method of binary complex system is formed based on Bayesian Network.Then,in view of the complexity of multistate systems,this paper proposes a compression inference algorithm for multistate systems.For the independent and non-independent multistate systems,this paper proposes multistate compression algorithms for various multistate node probability tables,which can realize effective compression for arbitrary multistate node probability tables.Based on the compressed node probability table obtained by the multistate compression algorithms and Bayesian Network,this paper proposes an independent and dependent multistate inference algorithm for the reliability modeling and analysis of the complex multistate systems,and also develops the construction rules of multistate intermediate variables involved in the reasoning process.Based on the multistate compression inference algorithm,the Bayesian Network reliability modeling and analysis methods for complex multistate systems are developed in this paper.Finally,the effectiveness of the proposed binary and multistate Bayesian Network compression inference algorithm is verified by numerical and engineering examples,which is applied to the reliability analysis of satellite systems and satellite attitude control system and the optimization of satellite reliability design.The proposed Bayesian Network compression inference algorithm,AgenaRisk software,the BNT toolbox of MATLAB are used to model and analyze the reliability of satellite system and satellite attitude control system.In the reliability design optimization of satellite system,the reliability analysis of satellite system is carried out based on the method proposed in this paper,and the overall design scheme of satellite is optimized by particle swarm optimization algorithm so that a better design scheme meeting the overall reliability design index of satellite is obtained.In this paper,the compression inference algorithm is studied deeply and comprehensively.The proposed binary and multistate compression inference algorithm can effectively solve the problem of state combination explosion in the reliability analysis of independent and dependent complex systems based on Bayesian Networks.In the overall design phase of satellite,the impact of uncertainty factors on the reliability of satellite is analyzed by compression reasoning algorithm,which guides the overall design optimization of satellite system.