Research and Application of Intelligent Fault Detection Method Based on Multi-scale Fusion and Enhancement

Fault is one of the key elements in the exploration and development of oil and gas reservoirs. The accurate prediction of faults is related to the accurate evaluation of oil and gas migration, reservoir prediction, and other work, which is of great significance. With the development of artificial intelligence technology, fault automatic detection technology based on deep learning algorithms has developed rapidly. In order to solve the problems of weak generalization ability and poor fault continuity of conventional deep learning algorithms for seismic data with different sampling rates, a new multi-scale prediction and fusion enhancement process has been proposed. Firstly, down sampling the input 3D seismic data on three dimensions is performed, and then fault detection is performed on the original data and its corresponding seismic data with three different scales. Subsequently, the prediction results from down sampling in three different directions are resampled back to the original scale, and the fusion results of the four fault predictions are obtained by taking the highest value. Finally, a fault scanning enhancement method based on matched filtering is applied to scan the fusion results along their possible fault strike and dip angles, suppress noise in fault prediction and improve the continuity of fault prediction, obtaining high signal-to-noise ratio and high continuity fault enhancement results. The field data results show that this method has obvious advantages in noise resistance, accuracy, and efficiency compared to traditional methods, and has broad application prospects in faults interpretation, development well position adjustment, and other work.