混合自适应量子粒子群优化算法

Hybrid adaptive quantum particle swarm optimization algorithm

为解决量子粒子群优化算法在迭代后期出现的种群多样性低导致的早熟收敛、陷入局部最优的问题,提出三点改进:(1)将收缩—扩张系数与适应度值联系起来,收缩—扩张系数会随着粒子的适应度值的改变而自适应调整;(2)使用差分策略更新粒子的随机位置,使得粒子向种群最优位置靠近;(3)粒子位置的更新加入Levy飞行策略,利用Levy飞行策略的偶尔长距离的跳跃,使得种群多样性增加,提高了跳出局部最优的能力,综合以上三点,提出一种混合自适应量子粒子群优化算法(HAQPSO).通过对比各个算法在6个典型函数的仿真测试结果表明:HAQPSO具有更好的全局收敛能力,且收敛精度、速度和稳定性都有明显的提升.

To solve the problem of quantum particle swarm optimization algorithm appears the poor diversity of population at the end of iteration,three improvements are proposed:(1)associating the contraction-expansion coefficient with the fitness value,the contraction-expansion coefficient will adjust adaptively with the change of the fitness value of the particle;(2)differential strategy is used to update the random location of particles which makes the particle approach the optimal position of the population;.(3)updating particle position with Levy flight strategy,Levy flight strategy′s occasional long jumps are utilized to increase the diversity of population and the ability of jump out of the local optimum.A hybrid adaptive quantum particle swarm optimization(HAQPSO)algorithm is proposed based on the above three points.By comparing the simulation result of 6 typical functions shows that:Improved quantum particle swarm optimization algorithm has better global convergence ability than the quantum particle swarm optimization algorithm,and the convergence precision,speed and stability are improved obviously.