基于有向加权复杂网络的自适应粒子群算法

摘要

标准粒子群算法在高维空间寻优迭代过程中存在易陷入局部最优和后期收敛速度慢的问题。引入复杂网络思想，提出一种基于有向加权复杂网络的自适应粒子群算法。该算法在粒子寻优的过程中引入有向动态网络进化机制，使粒子群的拓扑结构在入度服从幂律分布的条件下向无标度网络进化，同时根据粒子之间适应值的差值自适应调节动态学习因子的大小，使得粒子的飞行惯性在时间和空间上都是异质的，提高了粒子之间学习的多样性。仿真实验表明，该算法能够有效避免早熟问题，并且具有较快的收敛速度。%The disadvantages of particle swarm optimization(PSO)algorithm are that it is easy to fall into lo-cal optimum in high - dimensional space and has a low convergence rate in the iterative process. To deal with these problems,the complex network theory and an adaptive particle swarm optimization algorithm based on directed weighted complex network is proposed. At the same time,in the process of optimization, an evolutionary mechanism of the directed dynamic network is introduced and makes the topology of particles evolve into the scale - free network under the in - degree obeying power - law distribution. The value of the fitness’s subtraction of particles is employed to adaptively adjust the value of dynamic learning factor and en-sure the flying inertia of the particles is heterogeneous on time and on the space,improving the diversity of learning between particles. The results of simulation indicate that the proposed algorithm can improve the premature convergence problem and has a fast convergence rate.