Learning process patterns via multiple sequence alignment. (March 2022)

• Record Type: Journal Article
• Authors: Zheng, Chenglin; Chen, Xi; Zhang, Tong; Sahinidis, Nikolaos V.; Siirola, Jeffrey J.
• Is Part Of: Computers & chemical engineering. Volume 159(2022)
• Publication Date: 2022-03
• Subjects: Flowsheet comparison -- Process pattern -- Multiple sequence alignment -- Tree decomposition; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2022.107676
• Languages: English
• ISSNs: 0098-1354

Abstract

Highlights: Flowsheets carry a wealth of design knowledge. Learning from multiple process flowsheets can be done effectively with progressive string assignment algorithms. Learning from a single flowsheet can be facilitated by graph decomposition. Abstract: Chemical process flowsheets contain a wealth of hidden information. By recognizing similarities in process flowsheets, engineers can understand how existing processes were designed and improve the design and operations of new processes. In previous work, we introduced a systematic methodology for comparing two flowsheets. Our present work proposes a novel method based on multiple sequence alignment to explore general patterns among multiple process flowsheets simultaneously without having to conduct pairwise comparisons of a given number of flowsheets. Additionally, we propose a decomposition strategy to identify inner patterns within a single flowsheet. Several case studies demonstrate that the proposed methodology can compare multiple flowsheets simultaneously to determine general patterns and effectively identify patterns from a single flowsheet.