Prediction of Shale Oil Reservoir Brittleness Based on the Small-Sample Machine Learning Algorithm of SVM

Predicting the brittleness of shale oil reservoirs is a critical step in achieving efficient exploration and development, as it directly impacts the scale and complexity of fracture networks formed during hydraulic fracturing. A higher content of brittle minerals typically leads to larger, more intricate fractures, enhancing reservoir stimulation. Using the Jurassic Lianggaoshan Formation shale oil in the X1 well area of the Sichuan Basin as a case study, this research introduces a brittleness index prediction method based on the Support Vector Machine (SVM), a machine learning algorithm designed for small sample sizes. This approach addresses the limitations of conventional empirical formulas, which often suffer from low accuracy and specificity. By leveraging seismic pre-stack elastic parameter data, the method establishes a relationship model between the brittleness index and seismic elastic parameters through machine learning, enabling more precise predictions. Accurate prediction of shale reservoir brittleness index and compressibility evaluation have important guiding significance for shale oil exploration and development.