舰船主冷凝器多连片贝叶斯网络故障建模方法研究  

Research on fault modeling and simulation by multiple sectioned bayesian network of main condenser on ship

作　　者：许伟[1,2] 程刚[1,2] 耿江华[1,2] 黄林[1,2] 

机构地区：[1]海军工程大学舰船动力工程军队重点实验室,湖北武汉430033 [2]海军工程大学舰艇装备仿真技术研究室,湖北武汉430033

出　　处：《舰船科学技术》2016年第10期107-110,共4页Ship Science and Technology

基　　金：国家自然科学基金资助项目(51579242);湖北省自然科学基金资助项目(2013CFB440)

摘　　要：主冷凝器作为舰船蒸汽动力装置的主要设备运行中发生战损,故障模式复杂、不确定性较大,传统故障诊断方法难以有效解决。本文提出采用贝叶斯网络故障建模的方法解决这一难题,建立具有时间序列特性的动态多连片贝叶斯网络模型。通过实验分析表明模型准确可靠,不仅能够进行从故障原因到现象的正向推理,还能进行故障现象到原因的反向推理,可为蒸汽动力设备的故障诊断提供有效决策。As one of main steam power equipments on ship, main condenser makes battle damages in operation. Fault modes of main condenser are complex and uncertain, which traditional fault diagnosis method is difficult to solve. This paper presents a modeling method of Bayesian network to solve the problem and establish dynamic and contiguous Bayesian network model with time series character. The experiment shows that the model is accurate and reliable. The model can not only make forward inference from fault reasons to phenomenon, but also make backward inference from fault phenomenon to reasons, which can provide effective decision to fault diagnosis of steam power equipments.

关 键 词：主冷凝器 多连片贝叶斯网络 故障诊断 

分 类 号：TP393.092[自动化与计算机技术—计算机应用技术]