基于LPP特征空间重构的故障检测策略  

Fault Detection Strategy Based on the Feature Space Reconstruction of LPP

作　　者：张成[1] 赵丽颖 杨东昇 李元 ZHANG Cheng;ZHAO Liying;YANG Dongsheng;LI Yuan(Department of Science,Shenyang University of Chemical Technology,Shenyang 110142,China;Research Center for Technical Process Fault Diagnosis and Safety,Shenyang University of Chemical Technology,Shenyang 110142,China)

机构地区：[1]沈阳化工大学理学院,辽宁沈阳110142 [2]沈阳化工大学技术过程故障诊断与安全性研究中心,辽宁沈阳110142

出　　处：《沈阳化工大学学报》2023年第5期472-480,共9页Journal of Shenyang University of Chemical Technology

基　　金：国家自然科学基金资助项目(61490701,61673279);辽宁省自然科学基金项目(2019-MS-262);辽宁省教育厅基金项目(LJ2019013)。

摘　　要：针对多模态工业过程数据中存在方差差异显著的问题,提出了一种基于LPP特征空间重构的故障检测策略.首先,采用局部保持投影对过程数据进行降维处理,去除数据的冗余信息和噪声,降低计算复杂度;其次,将变分高斯混合模型应用于过程数据,确定操作模式的数量,并对每种模式下的数据进行聚类;再次,将每种模式下的数据利用所属模式的信息进行标准化处理,去除数据的多模态特征;最后,使用统计量T^(2)对过程进行监控.通过一个多模态数值例子和半导体蚀刻过程验证了所提方法的有效性.To solve the problem of fault detection in multimodal industrial process date with significant variance difference,a Feature Space Reconfiguration of Locality Preserving Projections(FSR-LPP)fault detection strategy.Firstly,the dimensionality reduction of process data is carried out by using LPP,which removes the redundant information and noise of the data,and the computational complexity is reduced.Then,applying a Variational Gaussian Mixture Model(VBGMM)to unlabeled process data,determine the number of operation modes,and cluster the data in each mode,secondly,the data in each mode is standardized using the information of the mode to remove the multimodal data features,finally,using statistics to monitor the process.The effectiveness of the proposed method was verified through a multimodal numerical example and semiconductor etching process.

关 键 词：多模态故障检测 局部保持投影 变分高斯混合模型 标准化 半导体蚀刻过程 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]