改进主成分分析的KNN故障检测研究  

KNN Fault Detection Based on Principal Component Research

作　　者：李元 白岩松 LI Yuan;BAI Yan-song(Shenyang University of Chemical Technology,Shenyang 110142,China)

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142

出　　处：《沈阳化工大学学报》2018年第4期366-371,共6页Journal of Shenyang University of Chemical Technology

基　　金：国家自然科学基金重点项目(61034006);国家自然科学基金面上项目(60774070;61174119);辽宁省教育厅科学研究一般项目(L2013155);辽宁省教育厅重点实验室基础研究项目(LZ2015059)

摘　　要：为提高过程故障检测的可靠性与准确性,提出改进的主成分分析的故障检测方法.首先通过K-means聚类算法将原始建模数据进行分类,然后在此基础上应用PCA在每个类中提取主成分,最后建立基于数据的KNN模型进行故障检测.研究结果应用于青霉素发酵过程,与传统的PCA、KNN算法进行比较,表明采用PC-KNN的故障检测结果更加准确,并且减少了数据量与计算时间,保证了过程的安全可靠运行.In order to improve the reliability and accuracy of process fault detection,an improved fault detection method of principal component analysis was proposed in this paper.Firstly,K-means clustering algorithm was used to classify the original modeling data.Then based on this,PCA was used to extract the principal components in each class.Finally,a data-based KNN model was established for fault detection.The results were applied to the penicillin fermentation process,and compared with the traditional PCA and KNN algorithms.It showed that the fault detection results using PC-KNN were more accurate,and the data amount and calculation time were reduced which ensured the safe and reliable operation of the process.

关 键 词：PCA K-MEANS KNN 故障检测 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]