A modular neural network-based population prediction strategy for evolutionary dynamic multi-objective optimization

This paper presents a novel population prediction algorithm based on modular neural network (PA-MNN) for handling dynamic multi-objective optimization. The proposed consists of three mechanisms. First, we set up (MNN) and train it with historical information. Some the initial solutions are generated by MNN when an environmental change is detected. Second, some predicted forward-looking center points. Finally, randomly to maintain diversity. With these mechanisms, new environment has been encountered before, will have same distribution characteristics as final that were obtained in last time. Because initialization mechanism does not need recent time, can also solve optimization problems dramatically irregularly changing Pareto set. tested variety test instances different difficulties. comparisons experimental results other state-of-the-art algorithms demonstrate promising dealing