A Non-Dominated Sorting Cooperative Co-Evolutionary Differential Evolution Algorithm for Multi-Objective Layout Optimization

Different from traditional multi-objective evolutionary algorithms (MOEAS), multi-objective cooperative co-evolutionary algorithms (MOCCEAs) divide the decision variables into several subproblems to optimize. Solutions of each subproblem are evaluated by complete solutions formed through combining representative solutions from all subproblems. Therefore, the combination of representative solutions is a key issue in MOCCEAs. To improve the capability of MOCCEAs to complex multi-objective optimization problems, we propose a non-dominated sorting cooperative co-evolutionary differential evolution algorithm (NSCCDE). The proposed NSCCDE uses an external archive for storing complete solutions to establish a new collaboration mechanism, which forms a complete solution by combining collaborators from each subpopulation as well as from the external archive. On the one hand, the external archive is updated continuously through non-dominated sorting of complete solutions, which is conducive to speeding up the convergence. On the other hand, the external archive evolves itself through spatial dispersal and mutation operation to increase the diversity. The performance of proposed NSCCDE is then evaluated on a suite of satellite module layout optimization problem. Experimental results demonstrate that the proposed algorithm outperforms NSCCGA and NSGA- II.