Abstract: A hierarchical Particle Swarm Optimization(PSO) algorithm is proposed in order to overcome the weak ability of local search and slowly converging speed of PSO algorithm in later period. In the algorithm, the global search and local search with standard particle swarms performs synchronously, and the local layer is decided by the better individual which is found in the global layer. A dynamic adaptive strategy of adjusting the local search space is adopted to avoid converging to local optimization, so the algorithm can retain the global convergence ability of PSO successfully. Experimental results on several benchmark functions indicate that the hierarchical PSO increases the speed of convergence and enhances the ability of local search.

Key words: hierarchical Particle Swarm Optimization(PSO), global search, local search

摘要： 针对粒子群优化算法存在进化后期局部搜索能力不强、收敛速度变慢的问题，提出一种分层粒子群优化算法。利用标准粒子群优化算法在整个搜索空间内进行全局搜索，由全局搜索获得的较优个体产生局部搜索区域，在局部区域内进行进一步搜索。为避免陷入局部最优，采用动态调整局部搜索区域的策略，保持算法的全局收敛性。通过典型测试函数计算表明，该算法的收敛速度和局部搜索能力有明显改善。

关键词: 分层粒子群优化, 全局搜索, 局部搜索

CLC Number: 

• TP301. 6