Full text loading...
-
Deep Learning Based 3D Velocity Field Nonlinear Multiple Regression
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 80th EAGE Conference and Exhibition 2018, Jun 2018, Volume 2018, p.1 - 5
Abstract
In this work we describe a Deep Neural Network workflow for 3D Velocity Field Nonlinear Multiple Regressive method, the algorithm has been designed to work even with a relatively small training set or large amount of features. By examining the time-velocity pair generated by supervised DNN, we arrive at that the workflow can well predict the velocity function all around the 3D seismic survey under the supervised seeds manual-picking time-velocity pairs. The use of this type of tool in the future will help geophysicist potentially reduce manual-picking control point density, reduce time spent on manual-picking and obtain the relatively accurate time-velocity pair through the Nonlinear Multiple Regressive prediction thus driving down the cost of processing a dataset. This method can also be used as an attribute interpolation tools such as seismic trace interpolation and recovery, well log interpolation and so on.