融合种群信息的并行和声搜索算法在结构优化中的应用  

Application of parallel harmony search algorithm with swarm information sharing in structural optimization

作　　者：曹鸿猷[1] 张冠宇 黄斌[1] 历明 CAO Hongyou;ZHANG Guanyu;HUANG Bin;LI Ming(School of Civil Engineering and Architecture,Wuhan University of Technology,Wuhan 430070,China)

机构地区：[1]武汉理工大学土木工程与建筑学院,湖北武汉430070

出　　处：《华中科技大学学报(自然科学版)》2024年第12期142-148,共7页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(51978545)。

摘　　要：针对标准和声搜索(HS)算法串行迭代特点及全局搜索能力不足的问题,提出了一种用于提高结构优化设计效率的具有并行计算能力且强化全局搜索性能的改进和声搜索算法.不同于标准和声搜索算法,改进算法构建了多个子和声记忆(SHMs),迭代过程中每个子和声记忆同时进化以实现并行运算并提高全局搜索能力.为了实现子和声记忆之间的信息传递和平衡算法的勘探与开发能力,提出一种基于粒子群优化(PSO)算法的种群信息共享策略的新和声生成算子.通过6个基准函数对比研究了改进算法较标准和声搜索及其他优化算法在勘探能力和开发能力方面的优势,最后将该算法用于600杆桁架优化问题,进一步评估其在大规模复杂结构优化问题中的性能和效率.优化结果表明:改进算法获得了比标准和声搜索算法更好和更稳定的最优解,并且因其所具有的并行能力,优化计算效率较标准和声搜索算法也有大幅提升.An improved harmony search algorithm with parallel computing capability was presented and global search performance to address the issues of standard harmony search algorithm was enhanced,namely their serial iterative nature and limited global search capability,for improving the efficiency of structural optimization designs.Unlike the standard harmony search,the improved algorithm constructed multiple sub-harmony memories.In the iteration process,the sub-harmony memories evolved simultaneously to achieve parallel computing and enhance global search capability.A new harmony generation operator,based on the swarm information sharing strategy of particle swarm optimization algorithm,was proposed to facilitate information exchange among different sub-harmony memories and balance the exploration and exploitation abilities of the improved algorithm.The improved algorithm was compared with the standard harmony search and other optimization algorithms in terms of exploration and exploitation capabilities for 6 benchmark functions.Finally,the improved algorithm was applied to 600 bar truss optimization problems to evaluate its efficiency and performance in large-scale,complex structural optimization problems.The optimization results show that the improved algorithm achieves better and more stable optimal solutions than the standard harmony search algorithm,and it significantly improves the computational efficiency due to its parallel capability compared to the standard harmony search algorithm.

关 键 词：和声搜索 结构优化 并行计算 种群信息共享 桁架 

分 类 号：TU323.4[建筑科学—结构工程]