贝叶斯网络用于屏蔽泵系统故障分析方法的研究  

Study on shielded pump system failure analysis method based on Bayesian network

作　　者：鲍依兰[1] 黄高峰[1] 佟立丽[1] 曹学武[1] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200240

出　　处：《核科学与工程》2012年第2期103-109,共7页Nuclear Science and Engineering

基　　金：国家"973"计划资助项目(2009CB724301)

摘　　要：将贝叶斯网络系统故障分析方法中,旨在提高故障分析中不确定逻辑关系知识表达和多故障状态的考虑。本文构建屏蔽泵系统故障的贝叶斯网络,对网络中故障参数进行贝叶斯方法学习,在已知新样本信息时更新网络参数,最后由贝叶斯网络推理,获得屏蔽泵系统的薄弱环节和最大可能的故障模式,及其发生概率。实例证明了贝叶斯网络可作为故障分析方法,具有更强的建模分析表达能力。This paper applies Bayesian network to the system aim to improve knowledge representation of the uncertainty lo in system failure analysis. A Bayesian network for shielded failure analysis, with an gic and multi-fault states pump failure analysis is presented, conducting fault parameter learning, updating Bayesian based on new samples. Finally, through the Bayesian network infe network parameter rence, vulnerability in this system, the largest possible failure modes, and the fault probability are obtained. The powerful ability of Bayesian network to analyze system fault is illustrated by examples.

关 键 词：贝叶斯网络 故障 不确定性 

分 类 号：TL364.4[核科学技术—核技术及应用]