Low‐Carbon Optimal Dispatch Strategy Incorporating Attention‐Based Dynamic Carbon Emission Factors of Thermal Power Units

Traditional scheduling methods fail to accurately capture the dynamic carbon emission characteristics of thermal power units across different operating stages, leading to inefficient unit selection, increased emissions, and higher operating costs. To address these challenges under low‐carbon constraints, this paper proposes a low‐carbon optimal scheduling strategy that integrates a dynamic carbon emission model into the decision‐making process. First, a nonlinear model of the dynamic carbon emission factor of thermal power units is constructed by combining spatiotemporal attention mechanisms and absolute position encoding. This model captures the complex relationships between unit type, fuel characteristics, and load levels, achieving high‐precision estimation of the carbon emission factor. Second, the obtained dynamic carbon emission factor is embedded into a power‐carbon coupled scheduling framework for heterogeneous generating units, thereby achieving coordinated operation of units under low‐carbon conditions and improving system‐level decision‐making. Finally, simulation results based on a provincial power grid in China show that, compared with traditional scheduling methods using static emission factors, the proposed strategy can reduce total operating costs by 2.69%, increase renewable energy utilization by 0.18%, and reduce carbon emissions by 3.35%, providing strong technical support for low‐carbon scheduling and carbon neutrality goals in power systems with high renewable energy penetration. © 2026 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.