多目标优化问题的粒子群算法仿真研究*

研究了一种用于求解多目标优化问题的粒子群算法(CMMOPSO)。该算法采用外部存档存储每一代产生的非劣解, 并且采用拥挤距离来维持外部存档规模, 同时提出一种新的全局最优粒子的选取策略(基于拥挤距离和收敛性距离)来提升粒子向Pareto前沿飞行的概率;为提升种群跳出局部最优解的能力, 以一定的概率对外部存档中粒子进行变异操作。通过典型的多目标测试函数对提出的算法进行检测, 结果表明，CMMOPSO算法在求解多目标问题上有一定的优势。因此, CMMOPSO可以作为求解多目标优化问题的有效算法。

Particle swarm optimizer simulation research of multi-objective optimization problems

This paper proposed a kind of particle swarm optimizer used in solving multi-objective optimization problem (CMMOPSO for short). In the CMMOPSO algorithm, used the external archive to store the non-dominated solutions at each iteration and adopted the crowding distance to keep the size of the external archive. Proposed a kind of novel strategy to select the global best particle(based on crowding distance and convergence distance) to improve probability of flying to Pareto front. At last, to improve the ability to escape from local optima, performed the mutation operation by the certain mutation probability. Conducted experiments on a set of classical benchmark functions. Simulation results show that the proposed algorithm good performance. Consequently, CMMOPSO can be used as an effective algorithm to solve multi-objective problems.