Abstract: As a key industry for carbon emissions reduction, the power sector must urgently strive to achieve low-carbon development. In traditional low-carbon dispatch research for wind power systems, carbon emissions are primar-ily linked to the generation side, with fixed carbon emission coefficients used for calculation. However, this ap-proach simplifies carbon emission behaviour and fails to account for the spatio-temporal variability in carbon emission quantification. To address this issue, considering the spatiotemporal variability of carbon emissions, this paper extends carbon emission research from the power generation side to the load side. Based on carbon emission flow theory and demand response theory, a two-stage low-carbon economic dispatch model for wind power systems is proposed. In the first stage, a source-side economic dispatch model for wind power systems is established, and artificial intelligence algorithms are used to solve the model and obtain dispatch strategies for each unit; In the second stage, based on the dispatch results from the first stage, carbon potential and emissions are calculated at each node using carbon emission flow theory, and a demand response low-carbon optimisation model is established. Finally, through an improved IEEE 33-node case study, the proposed method is validated to guide users in altering their electricity consumption behaviour, enhancing wind power utilisation, unlocking carbon reduction potential on the load side, and further reducing system carbon emissions