论文部分内容阅读
目前瞬变电磁测深资料的反演基本局限于一维,且多数是基于最小二乘法,实际上是将非线性问题做线性化处理,再利用迭代的方法逐步逼近最优解,所以在一定程度上会丢失许多细节信息,同时极度依赖初始模型,易陷入局部最优解,而非全局最优解.为提高瞬变电磁测深法的反演解释精度,本文以瞬变电磁测深的一维正演理论为基础,同时针对野外实测资料的特点,建立非线性反演问题的目标函数,考虑到现有非线性反演算法存在的易陷入局部优化和过学习等问题,首次引入了一种当前优化领域的十分新颖且性能优良的智能优化算法——ABC(人工蜂群)算法,来求解反演问题.ABC算法对求解多极值非线性问题优势明显,它不需要了解待求解问题特殊信息,只需要对问题进行优劣的比较,通过单独人工蜂的局部寻优的行为,最终将全局最优值突显出来,具有很快的收敛速度.通过在Matlab平台下对算法参数的详细试算研究,最终确定了一套最佳反演参数,并以二、三、四层的地电模型为例进行了验证.在野外实测资料的反演解释中,经过与实地钻孔资料及其他反演方法的对比分析,表明利用ABC算法建立的瞬变电磁测深一维反演系统具有明显的理论优势和实用价值.
At present, the inversion of TEM data is basically confined to one dimension, and most of them are based on the least squares method. In fact, the nonlinear problem is linearized and the iterative method is used to approximate the optimal solution step by step. Therefore, It will lose a lot of detail information, at the same time, it relies on the initial model too much and is apt to fall into the local optimal solution rather than the global optimal solution. In order to improve the interpretation accuracy of TEM, Based on the one-dimensional forward modeling theory and the characteristics of the field measured data, an objective function of the nonlinear inversion problem is established. Considering that the existing nonlinear inversion algorithms are apt to fall into the problems of local optimization and over-learning, the first introduction One of the most innovative and excellent intelligent optimization algorithms in the field of current optimization, the ABC (artificial bee) algorithm, solves the inversion problem.The algorithm has obvious advantages for solving multi-polar nonlinear problems, it does not need to know the solution to be solved The special information of the problem only needs to compare the advantages and disadvantages of the problem, and through the local optimal behavior of the artificial bee, the global optimal value is finally highlighted with a fast convergence speed. Finally, a set of optimal inversion parameters has been determined by the detailed calculation and analysis of the algorithm parameters under the Matlab platform, and the geoelectric model of two, three and four layers has been verified as an example. , The comparison between the data from borehole drilling and other inversion methods shows that the 1D inversion system of transient electromagnetic sounding established by ABC algorithm has obvious theoretical advantages and practical value.