Solving dynamic multi-objective problems with a new prediction-based optimization algorithm

This paper proposes a new dynamic multi-objective optimization algorithm by integrating fitting-based prediction (FBP) mechanism with regularity model-based estimation of distribution (RM-MEDA) for in changing environments. The prediction-based reaction aims to generate high-quality population when changes occur, which includes three subpopulations tracking the moving Pareto-optimal set effectively. first subpopulation is created simple linear model two different stepsizes. second consists some sampling individuals generated strategy. third employing recent strategy, generating effective search improving convergence and diversity. Experimental results on benchmark functions variety characteristics difficulties illustrate that proposed has competitive effectiveness compared state-of-the-art algorithms.