Efficient Multi-Objective Evolutionary Algorithm Based on Partial Order Ranking

This paper introduce a new and efficient multi-objective evolutionary algorithms (EMOEA) for solving unconstrained and constrained multi-objective optimal problems (MOPs).The novelty of our algorithm are: (1) defining a new partial-order relation by Better function,(2) the individuals are ranked by this partial-order relation,which is very different from the ranking described in ordinary MOEA,because the constrained conditions are merged into ranking,therefore when we evaluate an individual,we dont need to consider whether the individual is feasible,(3) adopting multi-parents crossover and dynamical mutate to keep the uniformity of search,(4) By using the theory of finite Markov Chain,the convergence properties of our algorithm are proved.We take several benchmark MO optimization problems to test our algorithm.The numerical experiments show that our algorithm is superior to other MOEAs in terms of the precision,the quantity,the distribution uniformity and diversity of solutions and the convergence rate of algorithm.