摘要
针对柴油机多源信息故障诊断中,由传感器采集的不确定性信息造成的融合误差问题,从多源传感器信息时效性角度,提出一种基于信息时效性机会窗口的动态贝叶斯网络故障诊断方法.首先,根据多源传感器不确定性信息的变化规律建立信息时效性机会窗口,计算得到目标状态偏离信息;其次,利用目标状态偏离信息动态调节贝叶斯网络观测节点信息效用,降低不确定性信息对融合误差的影响.R6105AZLD柴油机台架试验表明,引入该方法后故障诊断灵敏度增强,故障后验概率的对比差距提高到35%.
In order to solve the problem of fusion error caused by uncertain information collected by sensors in fault diagnosis of diesel engine with multi-source information,a fault diagnosis method of dynamic Bayesian network based on information timeliness opportunity window is presented from the point of multi-source sensor information timeliness.Firstly,the information timeliness opportunity window is established according to the change rule of uncertain information collected by multi-source sensor,and the deviation information of target state is calculated.Secondly,the information utility of Bayesian network observation nodes is dynamically adjusted by using deviation information of target state to reduce the influence of uncertain information on fusion error.The results of R6105AZLD diesel engine bench test show that the sensitivity of fault diagnosis is increased and the contrast difference of fault posterior probability is increased to 35% after the introduction of the method.
作者
王承远
徐久军
严志军
WANG Chengyuan;XU Jiujun;YAN Zhijun(Marine Engineering College,Dalian Maritime University,Dalian 116026,China)
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2019年第2期201-210,共10页
Journal of Dalian University of Technology
基金
国家自然科学基金资助项目(51509029)
辽宁省教育厅基金资助项目(L2015065)
中央高校基本科研业务费专项资金资助项目(3132015032)
关键词
动态贝叶斯网络
柴油机
机会窗口
目标状态偏离信息
dynamic Bayesian network
diesel engine
opportunity window
deviation information of target state