A prediction strategy based on decision variable analysis for dynamic Multi-objective Optimization

Abstract Many multi-objective optimization problems in reality are dynamic, requiring the algorithm to quickly track moving optima after environment changes. Therefore, response strategies often used dynamic algorithms find Pareto optimal. In this paper, we propose a hybrid prediction strategy based on classification of decision variables, which consists three steps. After detecting change, first step is analyze influence each variable individual convergence and distribution new environment. The second adopt different methods for variables. Finally, adaptive selection applied solution set generated steps, solutions with good diversity selected make initial population more adaptable can help converge while maintaining its diversity. experimental results performance show that proposed capable significantly improving compared five state-of-the-art evolutionary algorithms.