摘要
为了通过增强种群多样性提高对粒子全局寻优能力与寻优速度的平衡能力,该文提出一种自适应粒子群优化(APSO)算法。基于种群熵对标准粒子群优化(SPSO)算法的多样性进行了研究,给出一种离散式多样性评价策略。为了均衡SPSO算法的勘探和开发能力,该文分析了SPSO算法的惯性权值随多样性评价值变化而变化的动态函数关系,并将该函数关系融入APSO算法。为防止算法搜索后期过早陷入局部最优点,采用一种变异策略增强种群的多样性。仿真结果证明:APSO算法相比耗散粒子群优化(DPSO)算法,增加了对未探测空间的搜索能力,加速了粒子在整个解空间的寻优过程。在开发阶段,惯性权值随多样性的减少而递减,在勘探阶段,惯性权值随多样性的增加而增加。APSO算法较好地平衡了算法的全局搜索和局部细致搜索能力,可使粒子在较大范围空间内快速寻找到最优解所在的区域,并展开细致搜索。
In order to balance the ability between the global searching ability of particles and the optimization speed by enhancing population diversity,an adaptive particle swarm optimization(APSO)algorithm is proposed here.The diversity of standard particle swarm optimization(SPSO)algorithm is analyzed based on population entropy and a discrete estimate strategy of diversity is given.The dynamic function relationship between inertia weight and diversity of the SPSO algorithm is analyzed to balance the trade-off between exploration and exploitation and incorporated into the APSO algorithm.To avoid obtaining partial optimal solutions in the searching later stage,the APSO algorithm introduces a mutation strategy to enhance the diversity of the population.The simulation results show that compared with the dissipative particle swarm optimization(DPSO)algorithm,the APSO algorithm enhances the searching ability of the unexplored space and accelerates the searching process of the particles in the whole solution space.In the development phase,the inertia weight decreases with the decrease of the diversity;in the prospection phase,the inertia weight increases with the increase of the diversity.The APSO algorithm balances the global searching ability and the local careful searching ability,and the particles can search out the area of the optimum solution in a wider space,and develop careful searching.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2013年第3期344-349,共6页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(61202313
61261027
31260273)
国家科技支撑计划(2012BAH25F02)
江西省自然科学基金(20122BAB201044
20132BAB211020)
江西省教育厅科技项目(GJJ12642
GJJ12514
GJJ13637)
关键词
离散式多样性评价策略
粒子群优化
熵
变异策略
discrete estimate strategy of diversity
particle swarm optimization
entropy
mutation strategy
