The foresight methodology for transitional shale gas reservoirs prediction based on a knowledge graph

The sedimentary environment in the eastern Ordos area of China is complex and contains a large number of transitional facies environments. Yet, there are many characteristics such as large vertical lithology changes, complex lateral sedimentary environment changes, small monolith thickness and large organic matter content changes, which lead to large uncertainty in the prediction of favorable areas for transitional shale gas. As the intricate reservoirs continue to unfold, the conventional linear prediction methods find themselves facing an arduous path to meet the demands of development. The ever-evolving complexity of these reservoirs has outpaced the capabilities of these traditional approaches. It becomes apparent that a more comprehensive and adaptable approach is necessary to navigate the intricacies of these reservoirs and unlock their hidden potential. Therefore, we put forward a method of introducing knowledge graph into transitional shale gas reservoir prediction in the eastern Ordos region by using big data technology. Because artificial intelligence big data relies heavily on data tags, it is particularly important for the construction of tags. Firstly, a top-down knowledge graph in the field of reservoir prediction is constructed to determine the key parameters used in prediction, namely porosity, total organic carbon and brittleness index. Secondly, the decision tree knowledge graph optimization label is constructed in a bottom-up way. The key parameter of this prediction is the knowledge graph obtained according to the professional knowledge of reservoir prediction, so as to optimize the school label of U-net and reduce the workload of artificial judgment. The results of the combination of the two methods are applied to 11 wells in Daji area of Ordos, and the experimental results are consistent with the actual situation of the reservoir. Based on the foundation of theoretical knowledge, this method enhances the efficiency and accuracy of interpretation and evaluation. It provides fundamental and technical support for the selection of favorable areas in shale gas exploration and the evaluation of exploration and development prospects, particularly in transitional shale gas areas, which is innovative and advanced in the field. Presents a method of applying knowledge map to shale gas reservoir prediction.Making labels based on knowledge for deep learning training, which makes the theory more perfect.The method has been applied to 11 Wells in the Great Ji area of Ordos, and the experimental results agree with the actual reservoir conditions with an accuracy of 100%.