基于混沌映射的自适应退火型粒子群算法

An Adaptive Annealing Particle Swarm Optimization Based on Chaotic Mapping

粒子群优化算法是一种新型的群体智能算法,具有参数少、使用方便、效果好等优点,因而得到了广泛应用。为了改进粒子群算法的性能,在自适应粒子群算法和模拟退火粒子群算法的基础上提出基于混沌映射的自适应退火型粒子群算法,在局部最优解附近添加混沌扰动算子,使其具有突跳能力,进而提高全局搜索能力;将传统的惯性因子改为双重选择策略,不仅使惯性因子随着目标函数的变化而变化而且随着粒子当前位置与上一时刻位置的距离的变化而变化;采用线性递减加速因子来动态调整自身经验和群体经验在迭代中的作用。通过数值实验验证了改进算法的性能,结果表明改进的算法对于不同类型的函数的寻优能力要优于自适应粒子群算法和模拟退火粒子群算法。

Particle swarm optimization(PSO)is a new swarm intelligence algorithm.It has some advantages such as fewer parameters,easy implementation and good efficiency etc.such that obtains many applications.In order to improve the performance of PSO,based on adaptive particle swarm optimization and simulated annealing particle swarm optimization,a chaotic adaptive annealing particle swarm optimization based on chaotic mapping is proposed.A chaotic perturbation operator is added to near-optimal solutions to enhance the global search capability.A double selection strategy is adopted instead of the traditional inertia factor,which not only makes the inertia factor change with the change of objective function,but also with the change of distance between the current position and the previous one of the particle.The effect of self and group experience in iteration is dynamically adjusted by a linear decreasing accelerating factor.The performance of the improved algorithm is verified by numerical experiments.The results show that the improved algorithm is superior to adaptive PSO and simulated annealing PSO for different types of functions.