To address the problem that existing carbon emission factors calculation methods for nodes in power grids struggle to handle the output uncertainty caused by high-penetration wind and solar energy integration

a prediction method considering the uncertainty of combined wind-solar output is proposed. First

a joint probability distribution of wind-solar output is constructed

followed by scenario generation and reduction to obtain typical sequences of combined wind-solar output. On this basis

the carbon emission factors of each node are accurately calculated by combining power grid power flow and carbon emission flow theories

and a training sample set is established. Furthermore

a variant graph convolutional network is adopted to model the power grid topology and the electricity-carbon coupling relationship between nodes

enabling the rapid prediction of user-side node carbon emission factors in future periods solely based on source-load forecast data. Simulation results show that the proposed method maintains high accuracy even under high-penetration renewable energy scenarios

with a mean absolute error (MAE) of 0.029 t CO2/(MW·h) and a mean absolute percentage error (MAPE) of 3.9% for node carbon emission factor prediction

thus verifying its effectiveness.