基于直觉模糊贝叶斯网络多态系统可靠性分析

Reliability Analysis of Multi-state Systems Based on Intuitionistic Fuzzy Bayesian Networks

针对贝叶斯网络分析多态系统可靠性时各根节点不同状态的精确概率难以确定的问题,提出了一种基于直觉模糊贝叶斯网络的可靠性分析方法。该方法是将多位专家评估的各根节点不同状态发生的直觉模糊概率及各位专家权重通过直觉模糊加权平均算子求解出综合概率,由得分值函数和相应的数据处理获得各根节点不同状态发生的精确概率;通过贝叶斯网络计算叶节点不同状态发生的概率,以及叶节点处于不同状态时各根节点处于不同状态的概率和重要程度。仿真表明该方法能够精确地确定各根节点不同状态发生的概率,并能够有效提高贝叶斯网络处理不确定性问题的能力。

In view of the difficulty in determining the precision probability of each root node in different states when Bayesian Networks were employed to analyze multi-state systems reliability, a method based on intuitionstic fuzzy Bayesian networks is proposed, in which the comprehensive probability is figured out with intuitionistic fuzzy weigh-ted average operator on the basis of the intuitionistic probability and corresponding weight of each of different states of each root node given by each of several experts, the precision probability of each of several different states of each root node is obtained with the score function and the corresponding data processing,. The probability of leaf node in each of several different states is obtained with Bayesian networks, and the probability in each of several different states and the importance of each root node are calculated for the leaf node in each of several different states. The simulation results and their analysis show preliminarily that the method can accurately obtain the proba-bility of the root node in each of several different states, and can effectively improve the ability of Bayesian networks to deal with uncertainty problems.