基于多源信息融合及贝叶斯网络的小子样可靠性评估  

Small-sample reliability evaluation based on multi-source information fusion and Bayesian network

作　　者：赵科渊 徐格宁[1] 陆凤仪[1] 戚其松[1] 

机构地区：[1]太原科技大学机械工程学院,太原030024

出　　处：《起重运输机械》2018年第6期79-84,共6页Hoisting and Conveying Machinery

基　　金：国家基金(51275329)

摘　　要：针对故障树布尔逻辑难以描述事件之间的关系,在实际应用中可靠性数据获取不足的问题,提出一种新的基于多源信息融合方法和贝叶斯网络的小子样多态系统可靠性评估方法。以贝叶斯网络为基础,将ML-Ⅱ多源信息融合法应用于节点的可靠度确定中。同时,为避免不可用数据的影响,提出t检测对验前数据和样本数据进行相容性检验,解决了小子样系统中根节点可靠度难以确定的问题,提高了预测精度。基于铸造起重机主起升机构可靠性评估对提出的方法进行验证,结果表明所提出的方法能够很好地应用于铸造起重机起升机构的可靠性评估,可为提高起升机构可靠性提供支持和参考。Considering Boolean logic based fault tree is poor in describing the relationship between events and obtaining reliability data in practical applications, a new small-sample polymorphic system reliability evaluation method based on multi-source information fusion and Bayesian network is proposed. ML-II multi-source information fusion method is applied to determining the reliability of the nodes based on Bayesian network. In addition, to avoid the influence from unusable data, t-test is used for compatibility check on pre-test data and sample data, which eliminates the difficulties in determining the reliability of root nodes in small-sample system and improves the prediction accuracy. The method is further applied to the reliability evaluation of main hoisting mechanism of casting crane, proving that the method works great and can provide support and reference for improving the reliability of the hoisting mechanism.

关 键 词：贝叶斯网络 起升机构 T检验 多源信息融合 可靠性评估 

分 类 号：TH218[机械工程—机械制造及自动化]