A Weight Vector Bi-Objective Evolutionary Algorithm with Bi-criterion Evolution for Many-Objective Optimization

In the multi-objective optimization process, its main purpose is to obtain Pareto non-dominated solutions with well convergence and diversity, but in most cases the convergence and diversity of solutions are conflicting. In order to solve this problem, this paper proposes a new convergence and diver...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2022 IEEE 4th International Conference on Power, Intelligent Computing and Systems (ICPICS) s. 273 - 279
Hlavní autoři: Wang, Jiangtao, Chen, Hanning
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 29.07.2022
Témata:
bi-criterion evolution
Bi-goal evolution
Convergence
convergence and diversity estimation
Diversity methods
Estimation
Evolutionary computation
many-objective optimization
multi-objective optimization
Optimization
Pareto optimization
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In the multi-objective optimization process, its main purpose is to obtain Pareto non-dominated solutions with well convergence and diversity, but in most cases the convergence and diversity of solutions are conflicting. In order to solve this problem, this paper proposes a new convergence and diversity evaluation method to convert the multi-objective optimization problem into a bi-goal, more effectively evaluate the dominant relationship between individuals through the weight vectors, and then increase the selection pressure. We introduce the bi-criterion evolution can better balance the convergence and diversity of Pareto optimal solutions. Based on the proposed method, a new multi-objective optimization algorithm BiGE-BEW is proposed. Experimental results show that the proposed algorithm shows strong competitiveness in solving multi-objective optimization problems and greatly improves the performance of algorithms for solving multi-objective optimization problems.
DOI:10.1109/ICPICS55264.2022.9873807