人工神经网络专家系统在瞬变电磁法反演中的应用

The application of artificial neural network expert system to the transient electromagnetic method inversion

目前在瞬变电磁法反演方法中 ,人们使用较多的是电磁场反演拟合的方法。但往往存在以下不足 :首先这些反演方法理论复杂 ,不易掌握 ;其次它要求工作人员有丰富的工作经验 ,缺乏延续性 ,很难推广 ;再次数据处理量大 ,计算复杂。本文提出一个基于人工神经网络的专家系统。它避开了具体复杂的电磁场计算 ,只要经过适当的学习训练就能够解决复杂的实际问题 ;而且还具有学习记忆功能 ,使得瞬变电磁法的反演工作具有延续性和继承性。随着专家系统的不断完善 。

Fitting is used mostly in the transient electromagnetic method inversion at present.However there is a few of deficiencies as following.First,its complex theoretical system is no easy to be mastered;secondly,it is difficult to be popularized in the shortage of inheritance Therefore,scientific researcher is requested to have rich work experience;thirdly,computation complicated for its huge data package.In this paper,an expert system based on artificial neura network is put forward,which avoids detailed and knotted electromagnetic computation.Once afte appropriate training and learning,it can solve practical problem effectively.Furthermore,it memory capacity enhances the continuity and inheritance in the field of TEM inversion.A brigh outlook will be brought along with the consummation of this expert system.