Based on panel data from 30 provinces, municipalities, and autonomous regions in China from 2008 to 2019, considering the pivotal role of energy structure adjustment in carbon emission control, this study utilizes the Kaya identity and its extensions, combined with the SCAD (smoothly clipped absolute deviation) method for index selection, to construct a detailed assessment system and analyze the contribution of various elements to carbon reduction. By further incorporating a semi-parametric model, the dynamic relationship between energy structure adjustments and regional carbon emissions is revealed. The research indicates that controlling power generation and adjusting the energy structure are the two most effective methods for contributing to carbon reduction. Furthermore, optimizing the energy structure by increasing the proportion of hydropower significantly facilitates carbon reduction. The hydropower proportions of 8%, 38%, and 76% are identified as critical thresholds impacting carbon reduction significantly. Additionally, through an analysis of regional heterogeneity, differentiated strategies for adjusting energy structures are proposed.