中文摘要：

由于非线性特性地球物理反演一直以来都是一个比较困难的问题．近十年来，非线性反演方法如人工神经网络、遗传算法在地球物理数据解释中得到越来越多的应用，但目前基本仍限于一维反演问题．对于二维反问题。反演参数较多，神经网络反演运用较少．本文利用BP神经网络优化方法，实现了电阻率二维非线性反演．与传统线性化的迭代反演比较，神经网络反演能够克服传统方法的不足、获得更好的反演结果．

英文摘要：

Geophysical inversion is a very difficult problem due to its non-linear nature. In the past decade,non- linear inversion algorithms such as artificial neural network （NN） and genetic algorithm （GA） are ncreasingly used for the interpretation of using used using NN are limited to one-dimensional geophysical data. However, until now, most of geophysical inversions （1-D） models. As to 2-D inverse problems, NN inversion is hardly because of a large number of parameters. In this paper, 2-D resistivity non-linear inversion is the Back-Propagation （BP） neural network method. Compared to the traditional iterative inversion method through linearization, the neural network inversion is able to overcome disadvantages of the traditional inversion and obtain better results.