基于K-means最佳聚类的间歇过程故障诊断方法  

Intermittent Process Fault Diagnosis Method Based on K-means Optimal Clustering

作　　者：邵盟雅 吕锋[3] 宋学君[1,2] 郭振兴 SHAO Meng-ya;LV Feng;SONG Xue-juna;GUO Zhen-xing(College of Physics,Hebei Normal University,Shijiazhuang 050024,China;Hebei Key Laboratory of Photophysics Research and Application,Hebei Normal University,Shijiazhuang 050024,China;College of Career Technology,Hebei Normal University,Shijiazhuang 050024,China)

机构地区：[1]河北师范大学物理学院,石家庄050024 [2]河北师范大学河北省光物理研究与应用重点实验室,石家庄050024 [3]河北师范大学职业技术学院,石家庄050024

出　　处：《控制工程》2020年第9期1642-1648,共7页Control Engineering of China

基　　金：国家自然科学基金项目(61673160,60974063,61175059);河北省自然科学基金项目(F2018205102);河北省自然科学基金(F2018205178);河北省教育厅项目(ZD2016053);教育部“春晖计划”合作科研项目(Z20177023)。

摘　　要：在多时段间歇过程中,过程特性区域存在于两个相邻时段间的模糊过渡区域中。若不能准确将过渡时段提取出来,在采用多向主元分析(MPCA)方法进行故障诊断与分析过程中容易产生误判。为此,提出了一种基于K-means最佳聚类的间歇过程故障诊断方法。该方法首先在传统软时段划分中加入K-means最佳聚类方法,提高了子时段与过渡时段的区分度;然后利用MPCA诊断方法进行诊断分析。此方法提高了过渡时段的区分度,进而提升了故障诊断的精度。通过机床设备加工数据的仿真实验,证明了方法的有效性。The process characteristic region exists in the fuzzy transition region between two adjacent time periods in the multi-period intermittent process.If the transition period cannot be extracted accurately,misjudgment will easily occur during the process of fault diagnosis and analysis using multi-directional principal component analysis(MPCA).Therefore,an intermittent process fault diagnosis method which based on K-means optimal clustering is proposed.Firstly,the K-means optimal clustering method is added to the traditional soft time division to improve the discrimination between sub-period and transition period.On this basis,the MPCA diagnosis method is used to perform diagnostic analysis and simulation experiment on processing data with machine tool equipment.This method improves the discrimination of transition period,and further improves the accuracy of fault diagnosis.The results of the experiment confirmed the effectiveness of the method.

关 键 词：MPCA K-MEANS 最佳聚类 软时段划分 故障诊断 

分 类 号：TP274.5[自动化与计算机技术—检测技术与自动化装置]