基于最佳保留策略的遗传算法波阻抗反演  

Wave Impedance Inversion Based on Genetic Algorithm of Optimal Retention Strategy

作　　者：付小念 吴朝容[1] Fu Xiaonian;Wu Chaorong(College of Geophysics,Chengdu University of Technology,Chengdu Sichuan 610059,China)

机构地区：[1]成都理工大学地球物理学院,四川成都610059

出　　处：《工程地球物理学报》2018年第4期460-465,共6页Chinese Journal of Engineering Geophysics

摘　　要：基于褶积模型的遗传算法与传统最优化算法相比,它不依赖于目标函数梯度信息,整个计算过程只依赖个体适应度值,能同时对搜索空间中的多个解进行评估,具有搜索过程随机,全局搜索性能强的特点。本文放弃传统的赌轮选择方式,采取一种最佳保留策略选择方式,并用单道地震记录进行反演,对两种结果进行差异分析,可知改进遗传算法反演结果精度更高。在改进遗传算法中,子代保留了父代中最好个体的同时保证了种群基因的多样性,从而保证最终可以搜索到全局最优解,避免了标准遗传算法中的早熟收敛现象。Compared with the traditional optimization algorithm,the genetic algorithm based on the convolution model does not depend on the gradient information of the objective function.The whole calculation process only relies on the individual fitness value,and can evaluate multiple solutions simultaneously in the search space.It has the characteristics of random search process,strong global search performance.This paper abandons the traditional way of roulette wheel selection,adopts an optimal reservation strategy selection method,and uses the single-trace seismic record to carry on the inversion.Differentiating the two results shows that the improved genetic algorithm is more accurate.In the improved genetic algorithm,offspring retain the best individuals in their parents and meanwhile ensure the diversity of the population genes,so as to ensure that the global optimal solution can be searched finally,avoiding the premature convergence in the standard genetic algorithm.

关 键 词：遗传算法 波阻抗 最优化 反演 

分 类 号：P631.4[天文地球—地质矿产勘探]