Abstract: The nationally planned coal mining area in Shanxi Province is an important energy-rich and ecologically fragile area in the Yellow River Basin, and it has long been under the dual pressure of high-intensity coal mining and ecological environment protection. However, the spatiotemporal evolutionary characteristics and driving factors of its ecological quality remain unclear, which seriously restricts precise implementation of regional ecological restoration policies. Based on multi-source remote-sensing data from 2000 to 2020, in this study, a remote-sensing ecological index of the mining area (M-RSEI) was established, and the XGBoost-SHAP interpretable framework was used to reveal the spatiotemporal evolutionary characteristics and nonlinear driving mechanisms of its ecological quality. The results show that: (1) M-RSEI is more robust and applicable to the coal-mining area environment than the traditional RSEI model, and it can accurately characterize the dynamic evolutionary characteristics of the entire life cycle of "mining-expansion-restoration" in the mining area. (2) From 2000 to 2020, the ecological quality of the coal mining areas was generally at a medium level and showed a steady improvement trend. The average value of M-RSEI increased from 0.453 to 0.560, presenting a spatial pattern of "high in the southeast and low in the northwest" in terms of space. (3) The XGBoost-SHAP attribution analysis indicated that annual potential evapotranspiration, annual precipitation, and land-use type are the key factors affecting the spatial differentiation of ecological quality in the study area, and the influence of each factor on ecological quality has significant nonlinear response characteristics and threshold effects. Our conclusions can provide a scientific reference for resource-based regions to formulate different sustainable-development strategies and promote high-quality coordinated development of green mine construction and the ecological environment.