基于T-S模糊故障树和贝叶斯网络的多态液压系统可靠性分析

Reliability Analysis of Multi-state Hydraulic System Based on T-S Fuzzy Fault Tree and Bayesian Network

为解决T-S模糊故障树分析方法在液压系统可靠性分析过程中运算复杂和只能单向推理的问题,提出一种基于T-S模糊故障树与贝叶斯网络的多态系统可靠性分析方法。根据给出的T-S模糊故障树向贝叶斯网络转化的方法确定贝叶斯网络的模型结构与条件概率表,利用贝叶斯网络的推理算法计算顶事件发生概率、事件后验概率以及底事件重要度。该方法既能进行计算系统可靠性指标及重要度的前向推理,又能进行故障诊断的反向推理,而且计算公式简单。最后通过900t提梁机液压驱动系统工程实例验证了算法的可行性与有效性。

In order to solve the disadvantages of calculation complexity and single inference of T-- S fuzzy fault tree analysis method in hydraulic system reliability analysis, a reliability analysis method of multi--state hydraulic system based on T--S fault tree and Bayesian network was proposed. The model structure and conditional probability table of Bayesian network were determined according to the method of converting T--S fuzzy fault tree to Bayesian network~ the probability of top event, pos- terior probability of events and importance of basic events were computed by means of inference algo- rithms of Bayesian network. The method can calculate the system reliability indices and importance by forward inference, as well as carry out fault diagnosis by backward inference, and the calculation formula is simple. Finally, the feasibility and effectiveness of the proposed method were illustrated by the engineering example of hydraulic driven system of the 900t gantry--crane.