摘要
为了解决船用中高速发动机磨损故障诊断准确率偏低的问题,提出多源信息融合与贝叶斯网络集成的磨损故障诊断方法。利用贝叶斯参数估计算法进行多源故障征兆信息融合,通过大量发动机磨损故障实测数据,结合该领域专家知识,建构贝叶斯磨损故障诊断网络,并建立朴素贝叶斯分类器,简化融合结果,最终通过最大后验概率估计值识别磨损故障模式。经实际故障案例计算分析,验证了该诊断方法的有效性及网络模型建构的准确性。
To solve the problems of low accuracy rate in marine medium-high speed engine wear fault diagnosis, a method developed from Bayesian networks and multi-source information fusion was proposed.Firstly,Bayesian parameter estimation algorithm was applied to fuse multi-source wear fault information.Then,the Bayesian diagnosis model based on a large number of engine's wear-fault measured data and integrated with domain experts knowledge was constructed,and naive bayesian classifier was established to simplify the fusion result. Finally, by mean of calculating the maximum posterior probability estimation, the mode of engine wear fault was identified. The accuracy of model and the validity of wear fault diagnosis method were verified through actual wear fault cases' calculation and analysis,which suggests its great value of practical application has great value of practical application.
作者
王永坚
陈丹
戴乐阳
WANG Yongjian, CHEN Dan, DAI Leyang(School of Marine Engineering, Jimei University, Xiamen 361021, China)
出处
《集美大学学报(自然科学版)》
CAS
2018年第3期205-211,共7页
Journal of Jimei University:Natural Science
基金
福建省自然科学基金资助项目(2016J01251
2016J01311)
关键词
船用中高速发动机
贝叶斯网络
磨损故障诊断
多源信息融合
marine medium-high speed engine
Bayesian networks
wear fault diagnosis
multi-source information fusion