计算机应用研究2018，Vol.35Issue(4)：1130-1134,5.DOI:10.3969/j.issn.1001-3695.2018.04.037
基于PC-WKNN的多工况间歇过程故障检测方法研究
Study on PC-WKNN-based fault detection method in multimode batch process
摘要
Abstract
In the multimode batch process of complex industrial production,for the characteristics of lager variables,the center drift and the distinctly different modal variance of the multi-modal data,this paper proposed a PCA and weighed K-nearestneighbor fault detection method (PC-WKNN).First,it used PCA to extract the main features of the training data,and simplified data structure.Second,in the principal component space,it found the distance between the training data sample and the k nearest neighbor,and calculated the average distance between the K nearest neighbor and local nearest neighbors,took the reciprocal of the average distance as the distance weight,took the weighted distance as statistic D,D was able to eliminate the influence of center drift and distinctly difference of modal variance.Finally it determined the control line using t distribution.compared the calculation D statistical value of online data and control line,it realized on-line fault detection.Using single mode and multi-mode examples,as well as examples of penicillin data simulation experiments,it compared with the KNN method to verify the effectiveness of the method.关键词
主元分析/K近邻/多模态/故障检测Key words
PCA/KNN/multimode/fault detection分类
信息技术与安全科学引用本文复制引用
冯立伟,张成,李元..基于PC-WKNN的多工况间歇过程故障检测方法研究[J].计算机应用研究,2018,35(4):1130-1134,5.基金项目
国家自然科学基金重点资助项目(61490701,61673279) （61490701,61673279）
2015辽宁省教育厅基金资助一般项目(L2015432) （L2015432）
2015辽宁省自然科学基金资助项目(2015020164) （2015020164）
