کاربرد شبکه عصبی مصنوعی در مدل سازی دو بعدی معکوس داده های مقاومت ویژه مگنتوتلوریک

The magnetotelluric (MT) method is one of the geophysical techniques used in geophysical exploration mostly in geothermal, hydrocarbon, and the minerals deposits. In this study,   ۲D inversion of the MT resistivity data is performed by artificial neural network (ANN). To achieve the goal, a series of back propagation net has been developed and trained by artificial data of several models, which prepared use of a commercial software called EMIXMT۲D. ‏The capability of the trained network for each polarization mode (i.e. Transverse Electric,TE, or Transverse Magnetic,TM) has then been assessed by another series of data which have not been used for training. To simulate the field condition, ۳,۵, and ۱۰ percent of randomly distributed noise were added on some sets of data and their effects have been evaluated.  It has been found that for data set free of noise the average error of the ANN modeling is about ۶.۶۴% whereas, it reaches up to ۸.۹۷% for data with ۵ percent noise. These results indicate that the developed network is capable enough to inverse MT resistivity data and produce models such that their responses are relatively in closed agreement with the measured data.