摘要
针对粒子群算法容易陷入局部最优的缺点,在改变动态惯性权值的基础上,提出了一种动态迭代次数粒子群算法DIPSO(Dynamic Iterative Particle Swarm Optimization)。该算法根据每个周期内达到收敛的迭代次数不同,在一个周期内,当其和累积小于某个值时,就对其重新进行初始化,从而使算法具有动态的自适应。通过对几种典型测试函数的优化,结果表明,DIPSO算法的收敛速度明显优于PSO算法,收敛精度也有所提高。
Aiming at the shortage of particle swarm algorithm being trapped in local optimum easily,this paper presents a dynamic iterative particle swarm optimization algorithm. The convergent iterative time is different in every cycle, when the sum of the iterative time is less that x,the particle swarm will be initialized,the way makes the algorithm is adaptive. The experiments show the new algorithm is superior to PSO either in convergence speed or in convergence accuracy.
出处
《机械工程与自动化》
2008年第4期76-78,共3页
Mechanical Engineering & Automation
关键词
粒子群
迭代次数
自适应
particle swarm
iterative time
adaptability
