基于降维映射分析及贝叶斯分类器的故障诊断方法及其在仪表中的应用

Introduction of a Failure Diagnosis Method for Instruments Based on Dimensionality Reduction Mapping&Bayesian Classifier

提出了一种基于贝叶斯分类器及基于主成分分析(Principal Component Analysis,PCA)相结合的仪表故障诊断技术。该技术采用马氏距离研究表征仪表故障的关键复合参数,采用PCA研究复合参数与实际参数映射关系,从而建立从仪表故障-关键影响因素之间的对应关系,为仪表的故障诊断与故障溯源提供依据。利用某超声波流量计故障诊断数据库,对上述方法的故障诊断效果进行了验证,结果显示两型超声波流量计进行的故障智能诊断的故障正确识别率分别达到99.6%和96.5%,还对影响超声波流量计相关故障的关键影响因素进行了分析。

A novel failure diagnosing method based on the Bayesian classifier and the Principal Component Analysis(PCA)method is proposed in this paper.The main composite parameters that contribute to diverse failures are studied with the Mahalanobis distance method,while the mapping relationship between the composite parameters and actual influences is obtained with PCA dimension reducing method.The relation between actual influences and failure mechanisms of instruments is the obtained,which is helpful for failure diagnosing and reasoning.To verify the method proposed in this paper,an ultrasonic flowmeter fault diagnosis database is used and corresponding failure diagnosis is performed.With this method,the failure of the two types of ultrasonic flowmeters are intelligently diagnosed and good recognition rates with failure recognition rate 99.6%and 96.5%respectively,are obtained.The key influencing factors that influencing diverse failure modes of ultrasonic flowmeters are also analyzed.