An Evolutionary Membrane Algorithm Based on Competition Mechanism for Multi-objective Optimization Problems

Abstract

The increasing focuses on coordinated developments of society, economy and environment makes multi-objective optimization an important tool for solving real-world problems. Thus an evolutionary membrane algorithm based on competition mechanism (EMACM) is proposed in this paper, which incorporates advantages of the NSGA-II evolution and the distributed structure of the membrane computing. The communication process distinguished the membrane algorithm with other intelligent algorithms. To share information between evolved populations, best objects selected are communicated to the upper-layer membrane through the competition mechanism to eliminate dominated solutions. The skin membrane archives global best objects as elitists, and serves as guidance for inner evolution processes. Verified by test functions, the EMACM is able to find global solutions that are converged well, approximated closely to and covering as much as possible the real Pareto front, and distributed uniformly along the whole front. Compared with classical algorithms, the EMACM demonstrates better performances of convergence and diversity.