Stakeholder-oriented multi-objective process optimization based on an improved genetic algorithm

Yang Su, Saimeng Jin, Xiangping Zhang, Weifeng Shen, Mario R. Eden, Jingzheng Ren

Research output: Journal article publicationJournal articleAcademic researchpeer-review

24 Citations (Scopus)

Abstract

Multi-objective optimization (MOO) is frequently used to solve many practical problems of chemical processes but process designers only need a limited number of valuable solutions in the final results. In this study, an optimization strategy associated with an improved genetic algorithm was developed to search valuable solutions for stakeholders’ preference more purposefully. The algorithm was improved to reduce overlapping solutions as a result of the discrete variables in practical problems, and it allowed users to set a reference point or an angle associated with a reference point to make solutions converge into the preferred spaces. Three test functions and two practical problems were used to highlight that the proposed strategy could make designers optimize processes more efficiently. Especially, the angle-based algorithm could be more effective than the distance-based one on the tri-objective problems. Thus, the developed strategy is robust in the optimization of processes assisted with the designer's preference.

Original languageEnglish
Article number106618
JournalComputers and Chemical Engineering
Volume132
DOIs
Publication statusPublished - 4 Jan 2020

Keywords

  • Genetic algorithm
  • Multi-objective optimization
  • Preference
  • Process optimization

ASJC Scopus subject areas

  • Chemical Engineering(all)
  • Computer Science Applications

Cite this