摘要
根据多尺度逐次逼近思想,建立了多尺度逐次逼近退火遗传算法。该方法能有效地解决遗传算法中存在的收敛"早熟"问题。采用该方法对一个高速异常体进行了数值模拟试验,并对青藏高原东北缘阿尼玛卿缝合带东段上部地壳速度结构实际资料进行了处理,结果表明,多尺度逐次逼近退火遗传算法能够较好地应用于地震走时层析成像研究。
SAGA is a combined algorithm of Simulated Annealing （SA） and Genetic Algorithm （GA）. The method has the feature of higher searching efficiency than SA or GA alone, but can not radically solve the problem of convergence prematurity of GA. The multiscale SAGA algorithm, which is based on multiscale successive approximation, can effectively solve the problem. We used the method to reconstruct an anomalous high-velocity body in a numerical test model, and to study the upper crustal structure characteristics of the eastern segment of the A＇nyemaqen suture zone. Both test and practical application prove that the mulfiscale SAGA is a good tool for seismic tomography.
出处
《地震研究》
CSCD
北大核心
2008年第3期255-261,共7页
Journal of Seismological Research
基金
国家自然科学基金(40334040和40474049)
华北水利水电学院高层次人才科研启动项目资助
