Abstract:
Aiming at the disadvantages of traditional coherent volume attribute in fault prediction,such as false fault and poor noise resistance,this paper proposes a method for fault prediction using convolutional neural networks.First,construct a convolutional neural network model suitable for the fault characteristics of the actual work area,then train the network model using the partial frequency division seismic data and manually interpreted fault labels,and finally apply the trained model to the entire 3D seismic data for fault prediction.The actual seismic data prediction results show that the fault prediction results based on the convolutional neural network are in good agreement with the seismic data,and the fault detail description is better than the traditional seismic coherent volume attribute method.