多峰值全局优化的混合算法  

Hybrid algorithm for multi-peak global optimization

作　　者：刘文强[1] 张晓伟[2] 

机构地区：[1]西京学院数学教研室,西安710123 [2]西安电子科技大学应用数学系,西安710071

出　　处：《计算机工程与应用》2008年第33期43-45,47,共4页Computer Engineering and Applications

基　　金：国家自然科学基金No.60574075~~

摘　　要：根据遗传算法在较小的可行区域内一般有较好的求解结果这一事实,提出了一种混合算法。该算法先利用区间算法求解全局优化问题来得到包含所有最优解的小区间,随后运用遗传算法进行后续过程。算法能够有效缩小一个较大的可行区域空间,提供高适应值的初始种群,求出多峰值问题的全部最优解,提高算法的求解精度同时避免陷入局部最优。最后数值实验说明了算法的有效性。According to the fact that Genetic Algorithm can give generally the better solution in the smaller feasible domain,a hybrid algorithm is presented.The proposed algorithm firstly uses the interval method to obtain all small intervals for the global optimization,then employs the Genetic Algorithm to execute the later process.The hybrid algorithm can reduce efficiently the larger feasible domain,provides the initial population with high fitness,gives all optima of multi-peak optimization,improves the accuracy and avoid falling into local optimum.Finally,numerical experiments show that the algorithm works efficiently.

关 键 词：遗传算法 区间算法 全局优化 

分 类 号：O224[理学—运筹学与控制论] TP301[理学—数学]