With the double pressure of energy scarcity and environmental pollution, the demand for sustainable energy solutions is growing. In this paper, an improved quasi-linear demand response optimization method for virtual power plants based on conditional value-at-risk is proposed to improve the stability and reliability of wind and solar energy. The method integrates source, load and storage, and adopts day-ahead and intra-day scheduling to accurately deal with the forecast uncertainty and enhance the system flexibility and control accuracy. The day-ahead phase uses forecast data over long time scales to make plans, while the intra-day phase adjusts based on real-time data. The improved quasi-linear demand response reward mechanism and Shapley value income distribution model based on risk attitude are introduced to ensure that flexible load resources share demand response (DR) economic benefits fairly and share the risks brought by new energy fluctuations. Simulation results show that this method can effectively improve the scheduling efficiency and market competitiveness of VPP, and has wide application potential and economic benefits.