基于STM的人工鱼群并行算法的研究  

Parallel algorithm for artificial fish swarm based on STM model

作　　者：李双[1] 李文敬[1] 杨文[1] 徐照松[1] 杨荐[1] 

机构地区：[1]广西师范学院计算机与信息工程学院,广西南宁530023

出　　处：《广西大学学报（自然科学版）》2013年第6期1444-1450,共7页Journal of Guangxi University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(61163012);广西自然科学基金资助项目(2012GXNSFAA053218)

摘　　要：为解决人工鱼群算法在寻找食物源的过程中容易陷入局部最优,且搜索速度慢的问题,采用动态权衡因子策略减小算法的强随机性,并构建MPI+OpenMP+STM三级并行编程模型,将鱼群算法中的觅食、聚群、追尾行为规范为事务,实现多线程之间的并发执行。在复杂多峰函数优化问题上进行测试比较,结果表明,该算法有效提高了收敛速度和寻优性能并获得了较高的加速比。In order to prevent the artificial fish swarm algorithm from being trapped into local optima easily in the process of food source searching, and to improve the searching speed, this paper uses the dynamic weighting factor strategy to reduce the strong randomness of the algorithm, and construct three level of parallel programming model called MPI ＋ OpenMP＋ STM. This model normalizes the behaviors such as preying, following, swarming into translations and implement the concurrent exe- cution between multiple threads. The experiment is based on the testing comparison of complex multi-peaks function optimization issue. The results show that our method has improved converging speed and optimization performance efficiently and achieved higher speedup rations.

关 键 词：人工鱼群算法 动态权衡因子 软事务内存 并行算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]