A dynamic neighborhood balancing-based multi-objective particle swarm optimization for multi-modal problems. (1st November 2022)

• Record Type: Journal Article
• Authors: Gu, Qinghua; Wang, Qian; Chen, Lu; Li, Xiaoguang; Li, Xuexian
• Is Part Of: Expert systems with applications. Volume 205(2022)
• Publication Date: 2022-11-01
• Subjects: Multi-modal multi-objective problem -- Particle swarm optimization -- Dynamic neighborhood -- Mutation operator -- Adaptive parameters; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2022.117713
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: The purpose of this study is to solve multi-modal multi-objective problems. Using the adaptive parameter adjustment strategy to extend the search space. The dynamic neighborhood forming strategy can exchange the information between particles in time. The mutation operator is embedded to make the particle jump out of the local optimum. Abstract: To solve the multi-modal multi-objective optimization problems which may have two or more Pareto-optimal solutions with the same fitness value, a new multi-objective particle swarm optimizer with a dynamic neighborhood balancing mechanism (DNB-MOPSO) is proposed in this paper. First, an adaptive parameter adjustment strategy is developed to balance the local and global search, which takes the difference among niches into consideration. Second, according to evolutionary states, a mutation operator is alternatively utilized to construct new solutions for escaping from the local optima. Then, combined with current niching methods, the dynamic neighborhood reform strategy of non-overlapping regions is properly implemented, which can enhance the exploration and keep the population diversity in the decision space. To validate the effectiveness of the proposed algorithm, DNB-MOPSO is compared with the other five popular multi-objective optimization algorithms. It is also applied to solve a real-world problem. The experimental results show the superiority of the proposed algorithm, especially in locating more optimal solutions in the decision space while obtaining the well-distributed Pareto fronts.