In the context of high proportion of new energy grid connection, in order to ensure the safe and stable operation of the new power system, it is important to find the adjustment potential of various types of flexible resources. Therefore, a day-ahead and real-time two-stage collaborative optimal model of multi-class flexible resources was proposed. Firstly, the uncertainty of new energy power generation was simulated by scenario simulation and scenario reduction technology, and the types and advantages of flexible resources on the power side, user side and energy storage side were analyzed. Then, the two-stage optimal framework of day-ahead and real-time was designed, and the risk economy dispatching model was established. In the day-ahead stage, the start-stop plan and power generation plan were determined according to the predicted power of the new energy, and the power balance was achieved with the minimum operating cost. In the real-time stage, the power adjustment, call reserve capacity and demand response degree were determined according to the actual simulated power of the new energy, and the power rebalancing was achieved at the minimum expected cost. In order to characterize the uncertainty of new energy, conditional value at risk was introduced into the objective function to measure risk. Finally, the regional system was taken as an example for numerical analysis. The results show that multiple types of power supply could achieve mutual assistance and meet the system load demand jointly under the two-stage optimal system. The flexibility resources of the power side, user side, and energy storage side played a collaborative adjustment function to enhance the flexibility of the system. In addition, the two-stage risk economy dispatching model characterized the uncertainty through conditional risk value, which better balanced the economy and safety of system operation.