Title:An Adaptive Dual-Domain Prediction Strategy based on Second-order Derivatives for Dynamic Multi-Objective Optimization

Abstract:This paper addresses the problem of dynamic multi-objective optimization problems (DMOPs), by demonstrating new approaches to change prediction strategies within an evolutionary algorithm paradigm. Because the objectives of such problems change over time, the Pareto optimal set (PS) and Pareto optimal front (PF) are also dynamic. To accurately track the changing PS and PF in the decision and objective spaces, we propose a novel adaptive prediction strategy, which utilizes the concept of second-order derivatives adaptively in different domains. %to deal with DMOPs. Firstly, the changes in both the PS and the PF are considered in this paper, which makes the proposed a dual-domain based method. Firstly, we propose a dual-domain method, which takes into account changes in both the PS and the PF simultaneously. An adaptive strategy is adopted to self-adjust the proportion of the search space. Secondly, a second-order derivative prediction strategy is proposed to predicatively re-initialize the population. We compare the performance of the proposed algorithm against four other state-of-the-art algorithms from the literature, using DMOPs benchmark problems. Experimental results show that the proposed method outperforms the other algorithms on most of the test problems.