结合模式搜索法的混合MIMIC算法  

A Hybrid MIMIC Algorithm Combined with Pattern Search Algorithm

作　　者：张丹[1] 夏桂梅[1] 

机构地区：[1]太原科技大学应用科学学院,太原030024

出　　处：《太原科技大学学报》2016年第6期490-494,共5页Journal of Taiyuan University of Science and Technology

基　　金：太原科技大学校研究生教改项目(20133001)

摘　　要：分布估计算法是一种全局寻优能力较强而局部求精能力较弱的优化算法,为增强分布估计算法的局部寻优能力,将局部求精能力强,收敛速度快的模式搜索法引入到分布估计算法中,提出一种结合模式搜索法的混合MIMIC算法(PS-MIMIC).通过测试函数测试算法性能,并与标准MIMIC算法结果进行比较,结果表明该算法在解决优化问题时具有良好的性能,可以较快的寻找到最优值。Estimation of distribution algorithm is an algorithm with higher global optimization ability but lower par- tial refinement ability. To enhance the partial refinement capacity of this algorithm, this paper introduced pattern search algorithm which belongs strong local refinement ability and fast convergence rate to the estimation of distri- bution algorithm, and then proposed a hybrid MIMIC algorithm combined with pattern search algorithm. Through testing the performance of the algorithm with test function and comparing the results with the standard MIMIC algo- rithm, we conclude that the algorithm has good performance in dealing with optimization problem and the optimal value can be found rapidly.

关 键 词：分布估计算法 模式搜索法 MIMIC算法 

分 类 号：O221[理学—运筹学与控制论]