Abstract
The gas content characteristics in shale gas reservoirs, including volume and distribution, are critical for optimizing extraction and resource utilization. However, the relationship between gas content (Vg) and well logging parameters (e.g., porosity (POR), density (DEN), natural gamma (GR)) and geochemical parameters (e.g., total organic carbon (TOC), potassium-uranium-thorium ratio (U), organic matter maturity (Ro), resistivity (ρ)) remains poorly understood. Additionally, a specific gas content model for the southern Sichuan region has yet to be established. This study introduces a method combining Kernel Principal Component Analysis (KPCA) and Support Vector Regression (SVR) to predict Vg quantitatively. A cross-analysis of various parameters identified POR, TOC, U, ρ, PERM, and DEN as key factors influencing Vg. These were used to develop a shale gas content model based on KPCA-SVR, which was validated using data from three wells in the Changning area of southern Sichuan. The model showed minimal discrepancy between predicted and observed values, demonstrating its accuracy. This research contributes a high-precision, machine learning-based model for predicting shale gas content.