Novel competing evolutionary membrane algorithm based on multiple reference points for multi-objective optimization of ethylene cracking processes
Di Cong,Chunjing Shi,Yunfei Cui,Yongming Han,Zhiqiang Geng
DOI: https://doi.org/10.1016/j.chemolab.2021.104389
IF: 4.175
2021-10-01
Chemometrics and Intelligent Laboratory Systems
Abstract:<p>The increasing needs for optimizing more objectives simultaneously, like production amount, economic benefits, energy utilization efficiency or saving amount, and emission reduction, pushes the research development of multi-objective optimization, which becomes complex as the number of objectives increases. In order to solve the multi-objective optimization problem with more objectives, a reference point based competing evolutionary membrane algorithm is proposed, helping chemical processes realize products amount improvement, energy saving and emission reduction. The proposed algorithm makes selection in parallel membranes from the evolutionary population based on their distance to the reference points, and communicates selected solutions between membranes. Nearer to reference points, more possible to be selected. The competing communication rule designed helps the algorithm realize information sharing. All communicated elitists are archived in the skin membrane and realize bi-directional communication. Experiments on extensible test functions prove that the proposed algorithm is able to find converged solutions that are distributed uniformly and cover the real value ranges of objectives well. Applying the algorithm to realize the multi-objective optimization of the ethylene cracking process, the results achieves energy reduction of 777.62 tons of feed oil and emission reduction of 2387.29 tons of carbon dioxide for a cracking period.</p>
automation & control systems,computer science, artificial intelligence,instruments & instrumentation,statistics & probability,mathematics, interdisciplinary applications,chemistry, analytical