Neural Computing in Geophysics

本研究旨在开发和实现基于神经网络的反演方法，以应用于地球物理反演问题，特别是地震反演问题。这种方法的动机是改进沿地震线的弹性参数的映射。目前，对地震测线沿线的每个点进行这种类型的反演在计算上是昂贵的。然而，对于沿着一条线的选定数量的点，它可以用于在训练神经网络的同时推导出弹性属性，即得出连接权。在训练之后，来自中间表面点的地震道集可以被输入到网络中，以估计这些点的深度相关的弹性性质，从而得到完整的地图。这个问题最困难的部分是网络的训练。模拟退火法和遗传算法等全局优化方法将被用于网络训练，因为这些方法在地球物理反演中已被证明是成功的。

This research is to develop and implement inversion methods based on neural networks for application to geophysical inversion problems, particularly, seismic inversion. The motivation for this approach is to improve the mapping of elastic parameters along a seismic line. This type of inversion for each point along a seismic line is currently computationally expensive. However, for a selected number of points along a line, it can be used to derive the elastic properties while training a neural network, that is, deriving the connection weights. After training, seismic gathers from the intermediate surface points can be input to the network to estimate depth-dependent elastic properties for those points resulting in a complete map. The most difficult part of the problem is to train the network. Global optimization methods such as simulated annealing and genetic algorithms will be used for the network training since the methods have proven successful in geophysical inversion.