Intelligent prediction of favorable shale oil areas based on deep learning and batch normalization

The evaluation and prediction of favorable zones are essential for accurate shale oil resource assessments and achieving profitable development. The geological conditions of GL shale oil are complex, highly heterogeneous, and have unclear geophysical response characteristics, creating difficulties in classifying and evaluating enriched shale oil layers. This study uses productivity as the primary criterion for favorable-zone evaluation. By leveraging the nonlinear mapping capabilities of deep learning algorithms, we establish an intelligent prediction method for identifying favorable zones that functions based on well fracturing segment productivity as labeled data and seismic attributes as feature data. This method calculates the contribution of individual fracturing intervals based on oil-phase tracer concentrations, decomposing the total oil production of a single well into the productivity of each fracturing segment. The connection between fracturing segment productivity and seismic attributes depends on geodetic coordinates derived from borehole trajectory survey calculations. We analyze the seismic attributes that most strongly influence productivity to develop an intelligent prediction and evaluation model for favorable shale oil zones. This model integrates deep learning techniques and batch normalization algorithms to comprehensively explore the relationships between multiple seismic attributes and productivity. This proposed approach enables efficient and intelligent prediction of favorable zones in the shale oil G9 layer, achieving an accuracy rate of 82.61 % for favorable-zone predictions based on the productivity of completed wells. The results of this study may provide theoretical and technical support for optimizing well location deployment, well numbers, and horizontal segment length to effectively improve the drilling success rates of high-quality reservoirs. They also offer significant guidance and practical application value for the efficient development of shale oil.