机构地区: 遵义师范学院数学系
出 处: 《计算机应用》 2011年第1期82-84,117,共4页
摘 要: 为了保证粒子群算法求得的非劣解尽可能接近真实的Pareto前沿并保持多样性分布,提出一种基于交叉和变异的多目标粒子群算法(CMMOPSO)。在CMMOPSO中,首先识别Pareto前沿的稀疏部分包含的粒子,并对这些粒子进行交叉操作以增加多样性分布;接着对远离Pareto前沿的粒子进行变异操作,以提升粒子向真实的Pareto前沿飞行的概率。在基准函数的测试中,结果显示CMMOPSO比其他算法有更好的运行效果。 In order to minimize the distance of the Pareto front produced by Particle Swarm Optimization (PSO) with respect to the global Pareto front and maximize the spread of solutions found by PSO, a multi-objective particle swarm optimization based on crossover and mutation (CMMOPSO) : In the CMMOPSO, firstly, the number of particle in sparse part of Pareto front was defined and the crossover operator was employed to increase the diversity of the nondominated solutions; next, the mutation operation was used for the particles far away from Pareto front to improve the probability to fly to Pareto front. In benchmark functions, CMMOPSO achieves better solutions than other algorithms.
领 域: [自动化与计算机技术] [自动化与计算机技术]