Abstract: Realizing the early warning of carbon emissions in substations can effectively promote the low-carbon transformation of substations. Therefore, a hybrid multi-attribute carbon emission early warning method suitable for substations is proposed. Firstly, based on the characteristics of carbon emissions from existing substations, an early warning index system for substation carbon emissions is established from three dimensions: economic viability, energy efficiency and environment impact. Secondly, adopting the idea of maximum deviation sum of squares, an early warning model for substation carbon emissions is constructed based on combined weighting and grey relational analysis. The threshold intervals of carbon emission levels are established through the introduction of the 3σ principle, and the early warning levels of substation carbon emissions are determined according to the threshold intervals. Finally, by employing the above-mentioned early warning method, a study is conducted on the carbon emission early warning for ten domestic substation samples In addition, the main reasons of carbon emission risk of substations under different early warning levels are analyzed. The results demonstrate that the combined weighting method based on the sum of squares of the maximum deviation significantly improves the flexibility of the early warning model, and the grey relational analysis combined with the 3σ principle can comprehensively and accurately reflect the carbon emission level of each substation. The early warning results provide theoretical reference for substation managers in formulating carbon emission reduction plans.