文献类型：期刊文献

英文题名：Quality-related Fault Detection Method Based on LASSO-Orthogonal Canonical Correlation Analysis

作者：Song, Bing[1]; Jin, Yuting[1]; Guo, Tao[1]; Shi, Hongbo[1]; Tao, Yang[1]; Tan, Shuai[1]

机构：[1] East China University of Science and Technology, Key Laboratory of Smart Manufacturing in Energy Chemical Process, Ministry of Education, Shanghai, 200237, China

年份：2022

卷号：2022-July

起止页码：2591

外文期刊名：Chinese Control Conference, CCC

收录：EI(收录号：20224413019126)

基金：? This research is sponsoredby NationalNatural Science Foundation of China (No. 62073140, 62073141, 62103149), Shanghai Rising-Star Program(No. 21QA1401800), National Natural Science Foundation of Shanghai (No. l9ZR1473200), National Key Research and Development Program ofChina (2020YFC1522502, 2020YFC1522505),.

语种：英文

外文关键词：Correlation methods - Shrinkage

摘要：There are many process variables in modern process industries. But due to closed-loop control, not all process variables that deviate from the operating point will affect product quality. In this paper, a method called Least Absolute Shrinkage and Selection Operator-Orthogonal Canonical Correlation Analysis (LASSO-OCCA) is proposed, which can effectively detect fault and determine whether the fault will affect the product quality or not. Firstly, use LASSO to calculate the correlation coefficient between process variables and quality variables, and select process variables that are strongly correlated with quality variables according to the correlation coefficient. Secondly, the selected process variables and quality variables are modeled using OCCA to obtain quality-related and quality-unrelated statistics, and then control limits are determined. Finally, the proposed method is tested on a typical case and compared with the traditional method to verify its feasibility. ? 2022 Technical Committee on Control Theory, Chinese Association of Automation.