改进遗传算法的CSAMT一维反演

发布时间：2018-03-27 14:09

本文选题：可控源音频大地电磁　切入点：反演　出处：《湖南大学学报(自然科学版)》2017年04期

【摘要】：传统的可控源音频大地电磁法(CSAMT)反演方法属于线性或者局部线性,大都依赖初始模型.而遗传算法因其不依赖初始模型的特点而应用到CSAMT反演中.但是,标准的遗传算法存在早熟、局部收敛等问题.针对这些问题,对标准的遗传算法进行改进,采用排序法和最优保留策略相结合的选择算子,增强其种群多样性并保证其收敛性;采用父子竞争策略和自适应概率法相结合的交叉算子,能够防止好的父代个体被淘汰,又具有适应性.通过理论模型进行算法仿真验证,证明其有效性,说明改进遗传算法较标准遗传算法在CSAMT一维反演中有明显的改善.通过对实测数据进行反演,其结果与地质资料吻合,证明了其适应性.
[Abstract]:The traditional method of controllable source audio magnetotelluric method is linear or locally linear, and mostly depends on the initial model. Genetic algorithm is applied to CSAMT inversion because it does not depend on the initial model. The standard genetic algorithm has some problems, such as premature convergence and local convergence. Aiming at these problems, the standard genetic algorithm is improved, and the selection operator which combines the sorting method and the optimal retention strategy is adopted to enhance its population diversity and ensure its convergence. The crossover operator, which combines father and son competition strategy with adaptive probability method, can prevent the good parent from being eliminated and has adaptability. The algorithm is proved to be effective by the theoretical model simulation. It is shown that the improved genetic algorithm is much better than the standard genetic algorithm in CSAMT one-dimensional inversion. By inversion of the measured data, the results are in agreement with the geological data, which proves its adaptability.
【作者单位】：国家地球物理探测仪器工程技术研究中心(吉林大学)吉林大学仪器科学与电气工程学院;
【基金】：吉林省科技厅发展计划重点项目(20100349) 国家潜在油气资源产学研用合作创新研究项目(OSR-02)~~
【分类号】：P631.325;TP18

【相似文献】