摘要
针对工业生产过程故障检测模型不能及时更新的问题,提出了一种特征空间自适应k近邻(featurespace adaptive k-nearest neighbor,FS-AkNN)故障检测方法。首先利用主元分析对训练数据进行降维,构建特征空间,然后利用k最近邻方法建立故障检测模型。在过程监视过程中,提出了基于距离规则的自适应更新故障检测模型。通过一个数值例子和TE过程的仿真实验结果表明了该方法的有效性。
Failure models for production process usually cannot be updated in time. This study proposed a feature space adaptive k-nearest neighbor (FS-AkNN) fault detection method. Principal component analysis was used to reduce training data dimension and construct a feature space. A fault detection model was then established using a k-nearest neighbor method. Moreover, a distance-based rule adaptive update fault detection model was proposed for process monitoring. Simulation results of numerical example and the TE process showed the effectiveness of the method.
作者
郭小萍
徐月
李元
GUO Xiao-ping;XU Yue;LI Yuan(College of Information Engineering,Shenyang University of Chemical Technology,Shenyang 110142,China)
出处
《高校化学工程学报》
EI
CAS
CSCD
北大核心
2019年第2期453-461,共9页
Journal of Chemical Engineering of Chinese Universities
基金
国家自然科学基金(61673279)
国家自然科学基金重大项目(61490701)
辽宁省科学事业公益研究基金(2016001006)
关键词
自适应
k近邻(kNN)
主元分析
故障检测
adaptive
k-nearest neighbor (kNN)
principal component analysis
fault detection
