Abstract: Focusing on the forecasting and early warning of heavy precipitation in the key areas along the Sichuan-Xizang Railway, a heavy precipitation classification forecast model has been constructed for various subregions based on high spatiotemporal resolution Fengyun-4 (FY-4) satellite data and ERA5 reanalysis product for the summers of 2020—2024, combined with the Light Gradient Boosting Machine algorithm. The model's interpretability is analyzed using Shapley additive explanation (SHAP), and the distribution characteristics of key forecasting factors are analyzed. Results show that the model achieves a Critical Success Index (CSI) of 0.41 for heavy precipitation forecast in the key regions with a Probability Of Detection (POD) reaching 0.76, and demonstrates a strong forecasting capability. Regional model analysis indicates that the POD for heavy precipitation is 0.83 and the CSI is 0.53 in western Sichuan. The POD is 0.69 and the CSI is 0.33 in the southeastern region of the Qingzang Plateau, indicating significant regional differences in the forecasting. SHAP and statistical analysis show that heavy precipitation in the southeastern Qingzang Plateau is mainly dominated by satellite Brightness Temperature Difference (BTD) factors (such as BTD_6.25−7.1 and BTD_13.5−10.7) that reflect variations in mid- and upper-level water vapor and cloud top height, while thermal instability parameters (such as CAPE (Convective Available Potential Energy), K index) and low-level vertical motion are the main indicators for heavy precipitation in western Sichuan. 60 min prior to the occurrence of heavy precipitation, key satellite parameters and physical parameters already exhibited statistically significant differences. Satellite parameters begin to show notable evolutionary characteristics as early as 150 min prior to the precipitation, providing quantitative reference for early warning. Based on interpretable machine learning methods, it is possible to not only enhance the objective forecasting ability of heavy precipitation, but also deepen the understanding of regional heavy precipitation mechanisms. This study provides scientific support for disaster prevention and reduction along the key areas of the Sichuan-Xizang Railway.