Knowledge-guided multiobjective particle swarm optimization with fusion learning strategies

Abstract Multiobjective particle swarm optimization (MOPSO) algorithm faces the difficulty of prematurity and insufficient diversity due to selection inappropriate leaders inefficient evolution strategies. Therefore, circumvent rapid loss population premature convergence in MOPSO, this paper proposes a knowledge-guided multiobjective using fusion learning strategies (KGMOPSO), which an improved leadership strategy based on knowledge utilization is presented select appropriate global leader for improving ability algorithm. Furthermore, similarity between different individuals dynamically measured detect current population, diversity-enhanced proposed prevent diversity. Additionally, maximum minimum crowding distance employed obtain excellent nondominated solutions. The KGMOPSO evaluated by comparisons with existing state-of-the-art algorithms ZDT DTLZ test instances. Experimental results illustrate that superior other regard solution quality maintenance.