基于混合遗传算法的偶极子源定位的仿真计算  

Simulation of the Hybrid Genetic Algorithm in Dipoles' Localization

作　　者：刘祎[1] 江国泰[1] 

机构地区：[1]同济大学生命科学与技术学院,200092

出　　处：《上海生物医学工程》2003年第2期13-16,共4页Shanghai Journal of Biomedical Engineering

基　　金：国家自然科学基金(59937160)

摘　　要：首先提出一种新的混合遗传算法。在基于实数编码的基础上,通过嵌入一个最速下降算子,结合遗传算法和最速下降法两者的优点,并引入模拟退火的思想,即可改善原算法的局部搜索能力,又能进一步提高优化效率。为了验证算法的可行性,通过对脑电偶极子源定位的仿真计算,证明所提出的新算法与其它优化算法相比,在达到最优解的效率上有了明显的提高。A new hybrid genetic algorithm was introduced and discussed in detail on its basic principle,mathe-matic mechanism,characteristics and application. Based on real encoding, the hybrid genetic alogrithm integrated features of the standard genetic algorithm and the steepest descent method. It can effectively improve local search ability and achieve global optimal solutin in higher probability by adopted simulated annealing idea. The improved hybrid genetic alorithm was applied to dipoles localization in order to validate the rationality and robust of it. Result showed that convergence performance was greatly enhanced, compared with other traditional algorithms.

关 键 词：混合适传算法 偶极子源定位 仿真计算 模拟退火 最速下降法 

分 类 号：Q343[生物学—遗传学]