贝叶斯网络结构在线学习算法研究及应用.pdfVIP

ISSN 1009-3044 E-mail: eduf@ 与技术 年第 卷第 期（总第 期） Computer Knowledge And Technology 电脑知识 2008 4 8 35 Computer Knowledge And Technology 电脑知识与技术 Vol.4,No.8,December 2008, pp.2230-2232 Tel:+86-551-5690963 5690964 贝叶斯网络结构在线学习算法研究及应用 钱娜，宫义山 （沈阳工业大学，辽宁沈阳 110178） 摘要：现代大型机电设备的日趋复杂化和自动化导致设备故障现象和机理之间具有很大的不确定性，因此对故障诊断技术提出了 更高的要求。 针对汽车发动机的工作原理及其故障知识结构特征，基于贝叶斯网络理论，以机器学习中的增量学习为基础提出和研 究了在线式贝叶斯网络结构学习方法，并利用该方法对汽车发动机故障结构网络进行在线学习。 最后通过实验分析验证了在线式 贝叶斯网络故障诊断方法比起传统的贝叶斯网络方法以及专家系统方法，该方法在汽车发动机故障诊断结果中具有更高的准确性 和可靠性。 关键词：故障诊断；贝叶斯网络；汽车发动机；结构在线学习 中图分类号： 文献标识码： 文章编号： TP393 A 1009-3044(2008)35-2230-03 Application and Research of Online Learning Algorithm for Bayesian Networks Structure QIAN Na, GONG Yi-shan (Shenyang University of Technology, Shenyang 110178, China) Abstract: Nowadays, the complication and automatization of large mechanical equipment leads to the more uncertainty between fault phe- nomenon and mechanism. Higher requirements are put forward for fault diagnosis technology. Pointing to the working principle and fault structure character of Automobile Engine, based on the Bayesian Networks and the increment learning of Machine Learning, the method of online learning for Bayesian Networks structure is advanced and researched. The method is applied to online learning of the structure net- work of Automobile Engine Fault Diagnosis. In conclusion comparing with the traditional method of Bayesian Networks and Expert Sys- tem Method, we validate that the method of online learning Bayesi