摘要
提出一种基于邻域保持嵌和标准距离K近邻(neighborhood preserving embedding-standard distance k nearest neighbor rule, NPE-SDKNN)的故障检测方法来解决非线性和多模态问题。首先,使用邻域保持嵌入方法提取数据中的流形结构,对数据进行降维;其次,在低维空间计算每个样本的标准距离,将各模态间的数据调整到同一尺度;最后使用标准距离的统计量对故障进行检测。邻域保持嵌入能够解决非线性问题和降低计算复杂度,标准距离K近邻通过用标准距离替代原始距离,消除了数据的多模态特征,使用NPE-SDKNN方法进行故障检测,能够提高多模态数据的故障检测率。在田纳西伊斯曼过程运用NPE-SDKNN方法,结果表明,相对于K近邻、主元分析、邻域保持嵌入、标准距离K近邻方法,NPE-SDKNN具有更高的故障检测率。
A fault detection method based on neighborhood preserving embedding-standard distance k nearest neighbor rule (NPE-SDKNN) is proposed to deal with the problems of nonlinear and multi-mode.Firstly,the manifold structure of data using the neighborhood preserving embedding method is extracted and projected the data from the high-dimensional space to the low-dimensional space. Secondly, the standard distance of each sample in the low-dimensional space is calculated and adjusted the data between the modes to the same scale. Finally,the statistic of standard distances is used to detect the fault.The neighborhood preserving embedding method can solve the problem of nonlinear and reduce the computational complexity.At the same time, by replacing the original distance with the standard distance,the multi-model characteristics of the data are solved, and the fault detection rate of the multi-model data is improved. In this paper, the NPE-SDKNN method is used in the Tennessee Eastman process, and the results show that the NPE-SDKNN has a higher fault detection rate than the KNN, PCA, NPE, SD-KNN methods.
作者
张宁
李元
ZHANG Ning;LI Yuan(College of Information Engineering Shenyang university of Chemical Technology,Shenyang 110142,China;College of Science,Shenyang University of Chemical Technology,Shenyang110142,China)
出处
《自动化与仪器仪表》
2023年第3期39-44,共6页
Automation & Instrumentation
关键词
邻域保持嵌入
标准距离K近邻
非线性
多模态
故障检测
neighborhood preserving embedding
standard distance k nearest neighbor
nonlinear
multi-mode
fault detection
