The intensification of global climate change and the resulting environmental challenges have made carbon emission control a focal point of global attention. As one of the major sources of carbon emissions, the power sector plays a critical role in accurately quantifying CO₂ emissions, which is essential for formulating effective emission reduction policies and action plans. The average CO₂ emission factor of electricity (AEF), as a key parameter, is widely used in calculating indirect carbon emissions from purchased electricity in various industries. The International Energy Agency (IEA) reported an AEF of 0.6093 kgCO₂/kWh for China in 2021, while the Ministry of Ecology and Environment of China (MEE) officially reported a value of 0.5568 kg CO₂/kWh, resulting in a discrepancy of 9.43%. This study conducts an in-depth analysis of the calculation methodologies used by the MEE and IEA, comparing them from two critical dimensions: calculation formulas and data sources, to explore potential causes of the observed discrepancies. Differences in formula components include factors such as electricity trade, the allocation of emissions from combined heat and power (CHP) plants, and emissions from own energy use in power plants. Notably, the IEA’s inclusion of CHP allocation reduces its calculated emissions by 10.99%. Regarding data sources, this study focuses on total carbon emissions and total electricity generation, revealing that the IEA’s total carbon emissions exceed those of the MEE by 9.71%. This exploratory analysis of the discrepancies in China’s AEFs provides valuable insights and a foundational basis for further research.

China is a leading producer and consumer of coal, with coal being the dominant energy source. The accurate calculation of the mass carbon emission factor (EF_m) of coal is crucial as the carbon emissions from its combustion influence carbon emission assessment and policy formulation. However, discrepancies in EF_m values across documents, due to varying net calorific values (NCVs), carbon contents (CCs), and carbon oxidation factors (COFs), have posed challenges for enterprises in carbon emission calculations. By analyzing different coal types, it is found that for anthracite, the EF_m difference in different documents can reach 38.5%; for bituminous coal, it can reach 42.3%; and for lignite, it can reach 18.6%. These differences significantly affect carbon emission calculation accuracy, carbon allowance allocation fairness, and enterprise costs under the Carbon Border Adjustment Mechanism (CBAM). For instance, in 2023, the calculated carbon emissions of anthracite vary by over 300 million tons depending on the EF_m used. To address these issues, relevant departments should establish a unified EF_m release system, build a data sharing platform, and standardize enterprise testing standards to enhance the accuracy of carbon-related calculations and drive the low-carbon development of the coal industry.

The carbon content as received (C_ar) of coal is essential for the emission factor method in IPCC methodology. The traditional carbon measurement mechanism relies on detection equipment, resulting in significant detection costs. To reduce detection costs and provide precise predictions of C_ars even in the absence of measurements, this paper proposes a neural network combining MLP with an attention mechanism (MSA-Net). In this model, the Attention Module is proposed to extract important and potential features. The Skip-Connections are utilized for feature reuse. The Huber loss is used to reduce the error between predicted C_ar values and actual values. The experimental results show that when the input includes eight measured parameters, the MAPE of MSA-Net is only 0.83%, which is better than the state-of-the-art Gaussian Process Regression (GPR) method. MSA-Net exhibits better predictive performance compared to MLP, RNN, LSTM, and Transformer. Moreover, this article provides two measurement solutions for thermal power enterprises to reduce detection costs.

Global efforts to achieve carbon neutrality have highlighted the critical need for accurate carbon emissions accounting in the thermal power industry, which is a major contributor to greenhouse gas emissions worldwide. Standardized and high-precision direct measurement methods are essential to support emissions reduction strategies and carbon trading systems. Accurately measuring thermal power carbon emissions is key to “carbon peaking and carbon neutrality (dual carbon)” goals. The direct measurement method, via real-time monitoring of flue gas flow and CO₂ concentration, provides high-precision data and is an important industrial direction. This paper reviews its key technologies, compares technical adaptability, anti-interference, and cost-effectiveness, sorts out domestic and international standards, and analyzes core challenges (25~50% errors from complex flow fields, environmental interference, traceability issues, high costs, and standard-engineering gaps). It forecasts future directions (flow field rectification, optical tech localization, and digital twins). The results guide enterprises to optimize monitoring, reduce carbon trading data errors, support standard formulation, and help small-medium plants promote direct measurement, accelerating the carbon neutrality process of the thermal power industry.

This study systematically compares thermal power direct carbon emission data quality control systems in China, the EU, and the U.S., quantifying differences and proposing optimization strategies. Core results: (1) Regulatory frameworks: The EU’s “Directive-Regulation-Standard” system controls data uncertainty ≤ ±3%; the U.S. “Clean Air Act+40 CFR Parts 75/98” framework achieves Continuous Emission Monitoring System (CEMS) median accuracy ~2% but has regional gaps; China’s 2024 legalization of CEMS data leaves a 32% eastern–western installation gap (~90% vs.~58%). (2) Technical paths: The EU mandates CEMS for units > 20 MW (65% ultrasonic flowmeters); the U.S. uses CEMS for coal-fired (error ≤ 1%) and accounting for gas-fired units; China’s 70% S-type Pitot tubes have 17% error in complex flow fields, while pilot multi-channel ultrasonic flowmeters reach ±1.5%. (3) Mechanisms: The EU’s QAL1-QAL3+AST cuts uncertainty by 40–60%; the U.S. NIST calibration limits cross-plant deviation ≤ 1.5%; China’s big data boosts anomaly identification by 72% but lacks full-process control. (4) China-specific proposals: Mandate CEMS for units > 300 MW, build 3–5 national flue gas platforms by 2026, and offer 60% western equipment subsidies, supporting carbon data quality improvement and international mutual recognition.