Ensemble enhanced active learning mixture discriminant analysis model and its application for semi-supervised fault classification

作     者：Weijun WANG Yun WANG Jun WANG Xinyun FANG Yuchen HE Weijun WANG;Yun WANG;Jun WANG;Xinyun FANG;Yuchen HE

作者机构：Key Laboratory of Intelligent Manufacturing Quality Big Data Tracing and Analysis of Zhejiang ProvinceChina Jiliang UniversityHangzhou 310018China Mechanical and Electrical Engineering DepartmentZhejiang Tongji Vocational College of Science and TechnologyHangzhou 311231China Suzhou Institute of MetrologySuzhou 215004China 

出 版 物：《Frontiers of Information Technology & Electronic Engineering》 (信息与电子工程前沿（英文版）)

年 卷 期：2022年第23卷第12期

页      面：1814-1827页

核心收录：

学科分类：08[工学] 080401[工学-精密仪器及机械] 0804[工学-仪器科学与技术] 080402[工学-测试计量技术及仪器] 0838[工学-公安技术] 

基　　金：Project supported by the National Natural Science Foundation of China (No.61903352) the Natural Science Foundation of Zhejiang Province,China (No.LQ19F030007) the Project of Department of Education of Zhejiang Province,China(No.Y202044960) the China Postdoctoral Science Foundation(No.2020M671721) the Fundamental Research Funds for the Provincial Universities of Zhejiang,China (Nos.2021YW18,2021YW80,and 2022YW96) the Innovative Team Project of Fujian Institute of Metrology,China 

主　　题：Semi-supervised Active learning Ensemble learning Mixture discriminant analysis Fault classification 

摘      要：As an indispensable part of process monitoring, the performance of fault classification relies heavily on the sufficiency of process knowledge. However, data labels are always difficult to acquire because of the limited sampling condition or expensive laboratory analysis, which may lead to deterioration of classification *** handle this dilemma, a new semi-supervised fault classification strategy is performed in which enhanced active learning is employed to evaluate the value of each unlabeled sample with respect to a specific labeled *** samples with large values will serve as supplementary information for the training dataset. In addition,we introduce several reasonable indexes and criteria, and thus human labeling interference is greatly reduced. Finally,the fault classification effectiveness of the proposed method is evaluated using a numerical example and the Tennessee Eastman process.