基于改进遗传算法的可控源音频大地电磁法一维反演研究

发布时间：2018-01-17 13:12

本文关键词：基于改进遗传算法的可控源音频大地电磁法一维反演研究　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：可控源音频大地电磁法是一种重要的探测方法,它常应用于地质勘探与调查。可控源音频大地电磁法采用人工可控制发射源,具有可控制性、探测深度大、分辨率高等优点。反演过程是可控源音频大地电磁法的一个重要部分,反演结果的优劣直接影响探测结果,反演计算结果由反演方法决定,因此,反演方法的选择至关重要。经典的反演方法包括线性和局部线性的方法,这些方法易出现病态矩阵、依赖初始模型等问题。而遗传算法因其不依赖初始模型、具有较强的搜索能力、简单性、通用性等优点而被广泛应用。然而,标准的遗传算法也存在早熟和局部收敛等问题。因此,对标准的遗传算法进行优化是必要的。本文通过对标准遗传算法进行改进,增强了算法的有效性,克服了早熟和局部收敛等问题;通过具体模型及实测资料的反演计算,证明改进遗传算法具有一定的抗噪性能和实用性,能够应用到可控源音频大地电磁法的一维反演中去。本文围绕基于改进遗传算法的可控源音频大地电磁法一维反演的有效性、抗噪声性能和实用性进行研究,主要分为四个部分:1、研究可控源音频大地电磁法的基本理论和正演计算理论,对层状断面进行正演计算,为反演做好基础;2、研究遗传算法的基本理论,理解遗传操作每个过程的运行机理,找出标准遗传算法存在的不足之处;3、改进遗传算法,对于选择算子,采用排序法和最优个体保留策略相结合的方法,对于交叉算子,采用自适应概率法和父子竞争策略相结合的方法;4、给出具体的模型及其参数,进行改进遗传算法的反演计算,并与标准遗传算法相比较,验证改进遗传算法的有效性;加入随机噪声到改进遗传算法中,验证改进遗传算法具有抗噪声能力;根据实测的地质数据资料,进行基于改进遗传算法的可控源音频大地电磁法的一维反演仿真实验,验证改进遗传算法的实用性能。本文的改进算法能够应用到可控源音频大地电磁法的一维反演当中去,得到较好的效果,具有一定的抗噪声性能和实用性。
[Abstract]:Controlled source audio magnetotelluric method is an important detection method, which is often used in geological exploration and investigation. The inversion process is an important part of the controllable source audio magnetotelluric method. The inversion results directly affect the detection results, and the inversion results are determined by the inversion method. The classical inversion methods include linear and local linear methods, which are prone to ill-conditioned matrix, dependent on the initial model and so on. Genetic algorithm is not dependent on the initial model. Because of its strong search ability, simplicity and versatility, it is widely used. However, the standard genetic algorithm also has some problems such as premature convergence and local convergence. It is necessary to optimize the standard genetic algorithm. By improving the standard genetic algorithm, the effectiveness of the algorithm is enhanced, and the problems of premature and local convergence are overcome. Through the inversion calculation of the concrete model and the measured data, it is proved that the improved genetic algorithm has certain anti-noise performance and practicability. It can be applied to the one-dimensional inversion of controllable source audio magnetotelluric method. This paper focuses on the effectiveness, anti-noise performance and practicability of one-dimensional inversion of controllable source audio magnetotelluric method based on improved genetic algorithm. It is divided into four parts: 1. The basic theory and forward calculation theory of controllable source audio magnetotelluric method are studied. 2. Study the basic theory of genetic algorithm, understand the running mechanism of each process of genetic operation, find out the deficiency of standard genetic algorithm; 3. Improved genetic algorithm. For the selection operator, the sorting method is combined with the optimal individual retention strategy, and for the crossover operator, the adaptive probability method is combined with the parent-son competition strategy. 4. The concrete model and its parameters are given, the inversion calculation of the improved genetic algorithm is carried out, and compared with the standard genetic algorithm, the validity of the improved genetic algorithm is verified. Adding random noise to the improved genetic algorithm to verify that the improved genetic algorithm has anti-noise ability; Based on the measured geological data, the one-dimensional inversion simulation experiment of controllable source audio magnetotelluric method based on improved genetic algorithm is carried out. The improved genetic algorithm can be applied to the one-dimensional inversion of controllable source audio magnetotelluric method, and it has good anti-noise performance and practicability.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631.325;TP18

【相似文献】