Abstract: Coal-fired power plants are an important part of the global energy supply and also a major source of greenhouse gases such as carbon dioxide, exerting a profound impact on climate change. Accurately quantifying the carbon emissions from coal-fired power plants is a key technical challenge in achieving the “dual carbon” goals. Only through reliable emission quantification methods can the emission base be accurately grasped and effective emission reduction paths be formulated. In recent years, machine learning technology, with its powerful data modeling and prediction capabilities, has provided a new solution to this difficult problem. For this reason, this article briefly summarizes the development history and classification of machine learning technology, and systematically reviews the application progress and cutting-edge directions of machine learning technology in the research of carbon emissions from coal-fired power plants. Firstly, in view of the limitations of insufficient real-time performance of traditional accounting methods and high cost and low coverage of continuous Emission Monitoring Systems （CEMS）, this paper discusses how the Predictive Emission Monitoring System （PEMS） can achieve real-time prediction and data repair of carbon emissions through machine learning. Secondly, the electricity-carbon model based on power big data and machine learning technology were analyzed, and the application potential of Non-Intrusive Load Monitoring （NILM） in multi-device refined carbon emission decomposition was explored, significantly improving the interpretability of emission sources. Finally, by integrating satellite remote sensing and machine learning, wide-area inversion and outlier reconstruction of CO₂ emissions from coal-fired power plants were achieved, verifying the complementarity between satellite monitoring and ground-based inventories. Relevant studies have shown that machine learning has promoted the development of carbon emission monitoring towards real-time and intelligence through the integration of multi-dimensional technologies （PEMS-electricity-carbon model-satellite remote sensing）, but it still faces challenges such as model generalization and interpretability. In the future, it is necessary to strengthen the collaborative innovation of physical models and data-driven approaches, and build an integrated monitoring system of “air-space-ground” to provide technical support for the low-carbon transformation of global coal-fired power plants.