按概率突跳的改进微粒群优化算法

在基本微粒群优化算法(PSO)的社会心理学分析基础之上,提出了一种改进的微粒群优化算法,该算法中引入了一个新的参数,改写了原算法中粒子飞翔的速度公式,使粒子飞行时以一定概率在解空间内改变飞翔的距离和方向———突跳。对5个标准测试函数的优化结果表明,合理地选取新参数的大小,新算法能大幅度降低达到最优值所需要的进化代数,同时提高算法的收敛率,尤其是对高维复杂函数的优化效果更明显。

Modified particle swarm optimization via probabilistic leap

Based on the social psychology analysis of particle swarm optimization(PSO),a modified PSO was proposed.In this PSO a new parameter was introduced and the formula for flying velocity of the particle was modified, so that the event of a particle to leap stochastically to a different point can be taken into account.The simulations for 5 benchmark functions indicate that, when the new parameter is selected properly,the modified PSO can improve the search ability including efficiency and convergence dramatically,especially for high dimensional and complex functions.