Lexicographic Optimization Control Method Based Phase Recognition for Multiphase Batch Polymerization Reaction

To solve the problem that the correlation of variables varies with phase in multiphase batch polymerization reaction, this paper proposes an optimization method for lexicographic order switching based on phase recognition. This method uses kernel principal component analysis to analyze the correlation of each sampling time in a sliding time window, and analyze the phase of each batch of historical data. According to the analysis results, each batch is divided into individual phases, and the sub-phase data is used to identify the phase of real-time production. This method sets the corresponding lexicographic control target according to the characteristics of each phase of the process. According to the lexicographical order of real-time phase location, the optimal operation variable trajectory is obtained, so as to achieve the best optimization effect in phase, and ensure the safe production process and continuous and stable product quality. Copyright (C) 2022 The Authors.