【作者】 党振；

【导师】 姚成玉； 王智勇；

【作者基本信息】 燕山大学 ， 控制工程， 2012， 硕士

【摘要】 针对贝叶斯网络在可靠性分析中难以构造的不足，结合传统故障树和T-S模糊故障树，研究和完善了基于传统故障树转化的贝叶斯网络可靠性分析方法，提出了基于T-S模糊故障树转化的贝叶斯网络可靠性分析方法，提出了基于贝叶斯网络和PSO(粒子群优化)/μPSO算法的可靠性优化方法。将本文方法应用于分体式巷道运输车液压系统，为提高其可靠性提供理论依据。首先，系统地研究了贝叶斯网络的推理机制及建模特点，给出了由传统故障树构造贝叶斯网络的方法。利用贝叶斯网络推理不仅计算出顶事件的发生概率与基本事件的重要度，同时还可以得到基本事件的后验概率。将后验概率和故障分析处理成本作为故障属性考虑，提出了基于贝叶斯网络和逼近理想解排序法的多属性故障分析方法。其次，提出了基于T-S模糊故障树转化的贝叶斯网络可靠性分析方法。该方法利用贝叶斯网络推理，解决了T-S模糊故障树运算复杂和不能双向推理的不足。完成了基于传统故障树转化的贝叶斯网络可靠性分析方法和基于T-S模糊故障树转化的贝叶斯网络可靠性分析方法两种方法的对比分析。在此基础上，提出了基于贝叶斯网络和灰关联分析的多属性故障分析方法。再次，提出了基于贝叶斯网络和PSO/μPSO算法的可靠性优化方法。该方法是以贝叶斯网络可靠性分析方法为基础，构造了以基本事件故障概率为输入、顶事件故障概率为输出的故障函数关系，再结合费用等资源条件构造出两种可靠性优化模型，利用PSO算法与μPSO算法分别进行可靠性优化求解。这种方法扩宽了贝叶斯网络可靠性分析方法的应用范围。最后，利用上述方法，对分体式巷道运输车液压系统进行可靠性分析，并完成了两种可靠性优化设计方案。更多还原

【Abstract】 To solve shortage of Bayesian networks’difficulty to construct in reliability analysis,combining with traditional fault tree and T-S fuzzy fault tree, the reliability analysismethod of Bayesian networks built through traditional fault is studied and improved, thereliability analysis method of Bayesian networks built through T-S fuzzy fault tree ispresented, and the PSO(Particle Swarm Optimization)/μPSO algorithm is brought into, thereliability optimization method based on Bayesian networks and PSO algorithm ispresented. Then the presented methods sre applied to the coalmine roadway separatetransporter hydraulic system, and provides the theoretical basis to improve its reliability.Firstly, the inference mechanism and modelling features of the Bayesian networks areresearched systematically, and the construction method of Bayesian networks based ontraditional fault tree is given. The method based on Bayesian inference could not onlycalculates out the probability of top event and the importance of the basic events, but alsoobtains posterior probability of the basic events at the same time. And the posteriorprobability and fault analysis cost are considered as the fault attributes, multi-attributefault analysis method combined Bayesian networks and TOPSIS(Technique for OrderPreference by Similarity to an Ideal Solution) is presented.Secondly, to solve disadvantages of unidirectional reasoning and computationalcomplexity of T-S fuzzy fault tree analysis, the reliability analysis method based on T-Sfuzzy fault tree and Bayesian networks is presented. Comparative analysis betweenreliability analysis method based on traditional fault tree and reliability analysis methodbased on T-S fuzzy fault tree is completed. On this basis, multi-attribute fault analysismethod based on Bayesian networks and grey relation analysis is presented.Thirdly, the reliability optimization method based on Bayesian networks and PSOalgorithm is presented. This method is based on Bayesian networks reliability analysis,and constructs a function that the basic event fault probability as input and the top eventfault probability as output, then set up two reliability optimization models combining with the cost and other resources condition. According to the reliability optimization model,optimize the reliability of the system by PSO algorithm andμPSO algorithm. This methodwidened the application range of the Bayesian networks reliability analysis method.Finally, according to the presented method, the reliability analysis of coalmineroadway separate transporter’s hydraulic system is completed, and two reliabilityoptimization design scheme is completed.更多还原