文献类型：期刊文献

英文题名：Consistency-regularized dual-stage semi-supervised learning for prediction of attributes in chemical processes with incomplete datasets

作者：Liu, Xiaochen[1];Zhu, Jiannan[1];Li, Zhi[1];Cao, Yue[1];Zhao, Yunmeng[1];Yang, Minglei[1,2,3]

机构：[1]East China Univ Sci & Technol, Key Lab Smart Mfg Energy Chem Proc, Minist Educ, Shanghai 200237, Peoples R China;[2]East China Univ Sci & Technol, Engn Res Ctr Proc Syst Engn, Minist Educ, Shanghai, Peoples R China;[3]Huzhou Inst Ind Control Technol, Huzhou 313099, Peoples R China

年份：2025

卷号：318

外文期刊名：CHEMICAL ENGINEERING SCIENCE

收录：;EI(收录号：20252818734270);WOS:【SCI-EXPANDED(收录号:WOS:001529765600005)】；

基金：This work was supported by National Key Research and Development Program of China (2022YFB3305900) , National Natural Science Foundation of China (62293501, 62203173) , Major Program of Qingyuan Innovation Laboratory (Grant No. 00122002) and the Fundamental Research Funds for the Central Universities (222202517006) .

语种：英文

外文关键词：Semi-supervised learning; Consistency regularization; Continuous catalytic reforming; Incomplete data; Process systems engineering

摘要：In chemical processes, the prediction of product attributes is crucial for real-time optimization and control, however, incomplete datasets, due to various sampling frequencies, strongly affect the accuracy and efficiency of data-driven models. To address this, this paper proposes a semi-supervised learning method based on such datasets, namely Consistency-Regularized Dual-Stage Learning (CRDL). The method subdivides the dataset into high-frequency and low-frequency parts and identifies relationships between different types of output data. A two-stage model was designed. The first stage uses high-frequency data, and the second uses data reclassified based on low-frequency and output relationships. The method performance is evaluated by considering the error accumulation, the number of labeled data, the impact of the noise scale and the unsupervised loss coefficient. The effectiveness of CRDL on accuracy improvement is validated through the case study on Continuous Catalytic Reformer (CCR) process.