用退火遗传算法进行地震波阻抗反演

Wave impedance inversion via annealing and genetic algorithm

成都理工大学信息工程学院信物系,四川成都 610059

Department of Information Engineering and Geophysics, Chengdu University of Technology, Chengdu 610059, China

退火遗传算法是模拟退火算法(SA)和遗传算法(GA)相结合的优化算法。其主要思想是利用模拟退火算法中的热槽法(概率分布函数)产生GA的群体初值,这样可以避免产生一些无用的群体,其结果既节省计算机空间,又节省计算时间,并且最终容易得到最优解。SA和GA交替运行,对GA的每一演化代,用SA对其进行局部搜索来改进GA的性能。利用退火遗传算法对地震波阻抗进行反演,目标函数可以达到10^-5数量级,这就使退火遗传算法的非线性方法能够应用于地震波阻抗反演中。

Annealing and genetic algorithm is an optimizedalgorithm that integrate advantages of stimulated an-nealing (SA) and genetic algorithm (GA). The heat-bath method of SA is adopted to generate the initialcolonies of GA, which helps avoid useless colonies,save computer memory and computing time, and ob-tain optimal solution. The SA and GA run alternate-ly. At each evolutionary generation, the SA is adoptedto improve the local performance of the GA. The an-nealing and genetic algorithm can achieve an order ofprecision as high as 10^-5 in wave impedance inversion.