To enhance the tracking ability and anti-disturbance performance of the dispatching control of a power system with a concentrating solar power plant, and reduce the impact of uncertain factors such as wind and load power on the control effect, an optimal dispatch method that contains long-time scale receding horizon optimization layer and short-time scale dynamic adjustment layers is proposed. This is based on the optimization idea of model predictive control. In this model, the upper layer takes the optimal operating economy of the concentrating solar power plant system as the objective, and the dispatch plan in long time-scale is formulated by a multi-step receding horizon solution. While the lower layer takes the tracing and correcting scheduling plan in the upper layer as the objective, and the thermal energy storage system is introduced to further deal with the power fluctuations of wind power and load demand over a short-time scale. This is based on smoothing power fluctuation by receding horizon optimization on a short-time scale. An example based on the improved IEEE 39 bus system is analyzed. The results show that the proposed method can effectively suppress the short-time power fluctuation of wind and load and realize the economical operation of the system through the coordination of two layers.This work is supported by the National Natural Science Foundation of China (No. 51907164).