Multi-Objective Collaborative Optimization of Distillation Column Group Based on System Identification

In this paper, a multi-objective collaborative optimization (MOCO) strategy is proposed for making decisions on a distillation column group. Firstly, based on data preprocessing, the operating modes of the tower group are determined by use of the fuzzy C-means clustering method. Secondly, based on the proposed concept of a collaborative variable, the discrete state-space model of the main towers are constructed by the subspace identification method. Then, a MOCO optimization model is designed for the ethylene plant. Finally, NSGA-III is used to solve the optimization problem. An analysis of a Pareto-optimal frontier and population is carried out. To illustrate the superiority of the proposed strategy, the results are compared with historical data and the appealing operation area is finally obtained.