一种惯性权重与种群多样性协同调整的二进制粒子群优化算法

Binary Particle Swarm Optimization Algorithm for Coordinated Adjustment of Population Diversity and Inertia Weight

针对二进制粒子群算法惯性权重和种群多样性不能随粒子群进化状态变化而动态协同调整,易造成后期收敛性较差陷入局部最优的缺点,提出一种惯性权重与种群多样性协同调整的二进制粒子群算法(CBPSO),首先使用混沌函数初始化种群,其次根据每个粒子与最优粒子之间海明距离均值与当前迭代状态共同调整权重值,再次根据海明距离均值动态调整种群多样性,最后根据调整后的种群多样性在下次迭代中计算新的海明距离均值及对应的惯性权重值.通过对常用的基准函数进行不同维度下的仿真实验,实验结果证明:在相同迭代次数等条件下,该算法具有较强的动态搜索能力和种群多样性调整能力,比同类算法具有更好的准确率和鲁棒性.

For the inertia weight and population diversity of binary particle swarm optimization （BPSO） cannot dynamically coordina- ted adjust with the evolutionary state of particle swarm, which be easy to cause poor convergence performance and fall into local opti- mum at later period. A new binary particle swarm algorithm （ CBPSO ） based on coordinated adjustment of population diversity and in- ertia weight is proposed. Firstly, this algorithm initialized population with chaotic function;secondly,it adjusted the inertia weight value according to the mean of Hamming distance between each particle and the optimal particle and the current iterative state; and then the algorithm dynamically adjust population diversity according to the Hamming distance; finally, the new Hamming distance and the asso- ciated inertia weight values will be calculated according to the adjusted population diversity in the next iteration. Through the simula- tion experiments with commonly benchmark function under the different dimensions, the results show that this algorithm has strong dy- namic searchinj~ ability and population diversity adiustment ability in the same iterations, and has better accuracy and robustness.