引入改进鸽群搜索算子的粒子群优化算法.PDF

第31 卷 第10 期 模式识别与人工智能 Vol． 31 No． 10 2018 年10 月 Pattern Ｒecognition and Artificial Intelligence Oct． 2018 引入改进鸽群搜索算子的粒子群优化算法 马 龙1 卢才武1 顾清华1 阮顺领1 摘 要 随着迭代计算过程的推进，标准粒子群算法后期容易出现收敛速度较慢、精度较低、早熟及开发探索能力 较弱等问题． 针对上述不足，文中提出引入改进鸽群搜索算子的粒子群优化算法，采用Beta 反向学习策略进行种群 的初始化，实现种群粒子分布的多样性． 首先采用线性和非线性变异策略改进鸽群算法中的地图罗盘算子，提高鸽 群算法的开发和探索能力． 然后采用改进后的组合优化鸽群算子更新粒子群中粒子位置和速度，提高收敛速度和 计算精度，避免算法陷入局部最优． 实验表明，文中算法提高收敛计算速度，精度达到函数设定的理想值． 关键词 改进鸽群搜索算子，粒子群优化，Beta 分布函数，反向学习策略 引用格式 马龙，卢才武，顾清华，阮顺领． 引入改进鸽群搜索算子的粒子群优化算法． 模式识别与人工智能， 2018 ，31 (10):909 －920 ． DOI 10 ． 16451/j ． cnki． issn1003-6059 ． 201810005 中图法分类号 TP 18 Particle Swarm Optimization with Search Operator of Improved Pigeon-Inspired Algorithm MA Long1 ，LU Caiwu1 ，GU Qinghua1 ，ＲUAN Shunling1 ABSTＲACT The standard particle swarm optimization is easy to have problems of low convergence speed and precision ，prematurity and poor exploring ability during later period． Aiming at these problems ，an optimized particle swarm optimization based on improved pigeon search operator is proposed． Population initialization is determined by Beta opposition-based learning strategy ，and the diversity of population distribution is realized． The map compass operator is improved by the linear and nonlinear mutation strategy to improve the development and the exploration ability of the pigeon-inspired algorithm． Then ，the location and the speed are updated by the improved combination optimization operator to speed up the convergence ，enhance the precision and avoid falling into local optimal solution