Against the backdrop of the "dual carbon" target, the power sector has become an important part of carbon reduction. Virtual power plants (VPP) can further improve their overall efficiency by integrating and aggregating distributed resources to participate in the carbon market. However, the uncertainty of distributed new energy output poses many challenges for their operation and management. Therefore, on the basis of using the scenario generation and scenario reduction method based on Latin hypercubic sampling to deal with the uncertainty problem of wind power and photovoltaic output of distributed energy, the VPP, which aggregates multiple units and takes into account the user-side demand response, participates in electric energy market as well as the carbon market as a whole, and the optimal scheduling model with the minimum total cost of the VPP is constructed, which is finally solved by using the improved gray wolf optimization algorithm. Through comparative analysis of different scenarios, it can be concluded that the existence of carbon market and demand response enhances the consumption of clean energy such as wind power and photovoltaic, and reduces greenhouse gas emissions, and reduces the operating cost of the VPPs, and takes into account its economy and environmental protection.