Schemes analysis on immune multiobjective optimization in dynamic environments

In this paper, a parallel immune algorithm (PIA) inspired by artificial immune system is proposed for solving dynamic multiobjective problems (DMOPs). The population is divided into nondominated subpopulation and dominated sub-population. One focuses on exploring Pareto-optimal set through immune clone and hypermutation, while another is exploited and improved via the simulated binary crossover and polynomial mutation. To improve the tracking ability, the change detection and change reaction are developed in PIA. In numerical experiments, the effect of change detection threshold and individual updating ratio on the performance of PIA is interpreted and analyzed from different perspectives and further possible research direction is discussed.