Horizon and Fault CNN Identification Technique for Coal Field 3D Seismic Data

The identification of horizons and faults is an important aspect of 3D seismic structural interpretation. Conventional horizon and fault interpretation is mainly achieved through human-computer interaction, and the interpretation results are not only influenced by human factors, but also have low interpretation efficiency. In this paper, a convolutional neural network (CNN) technique for simultaneous automatic identification of horizons and faults is proposed. It uses structural geological modeling methods to build a labeled data cube, complete CNN model training, and realize simultaneous prediction of horizons and faults. Taking real coal field 3D seismic data as a case, the CNN method was used for automatic identification of horizons and faults. The results showed that this method can simultaneously identify horizons and faults, not only improving interpretation efficiency but also improving interpretation accuracy. It will become a new tool for coal field 3D seismic data interpretation.