The Concentrating Solar Power and Photovoltaic Power (CSP-PV) system have a reactive power control problem. To address this, a multi-time scale reactive power optimal control strategy based on MPC is proposed. Based on the day-ahead optimal reactive power of a synchronous generator and photovoltaic inverter, the bus voltage is controlled by dynamic reactive power compensation equipment within the intra-day time scale. This adopts the idea of rolling optimization and correcting control based on MPC. The voltage prediction model based on sensitivity is used to predict the voltage states at multiple moments in the future. On this basis, an intra-day rolling optimization model is established with minimum deviation of the predicted voltage control at multiple future moments as the optimization objective to obtain the reactive power control plan of the dynamic reactive power compensation equipment. The model predictive control of the in-day-reactive power and voltage is completed by feedback correction of the voltage control deviation. Taking the CSP-PV combined power generation system in the DUN Huang area of China as a simulation example, by comparing with the traditional multi-time scale reactive power optimization control strategy, the feasibility and effectiveness of the proposed control strategy in improving the system bus voltage and the stability of the PCC bus voltage for concentrating solar power station and photovoltaic power station are verified. This work is supported by Science and Technology Project of State Grid Corporation of China (No. SGGSKY00FJJS 1800140).