The multi-objective optimization of esterification process based on improved NSGA-III algorithm

The esterification in polymerization process is a key step in polyester fiber production. There are three optimization objectives in the process of esterification, including rate of esterification (E
s
), average molecular weight (M
n
), and diethylene glycol content percentage (W
t
), which are in conflict with each other. However, currently, in many literatures they all considered the two performance indicators which can not show entirely the process of PET polymerization process. Then, this paper firstly performs a three-objective optimization problem in esterification process with NSGA-III. Additional, the existing multi-objective algorithm NSGA-III can provide a set of solutions, but the diversity of the solutions is poor. Therefore, an improved NSGA-III algorithm, called PNSGA-III, is proposed in this paper to deal with the three-objective optimization problem in esterification process. The PNSGA-III has been compared with original NSGA-III on 9 widely used benchmark test problems WFG1 to WFG9, and the three-objective optimization problem in esterification process. Experimental results indicate that PNSGA-III outperforms NSGA-III in terms of IGD metric and HV metric. Meanwhile, PNSGA-III reaches a better diversity in the esterification problem.