The continuous development of microgrid technology has led to the increasing diversification of microgrid energy storage systems. In view of how the multi energy storage systems operate efficiently and economically, has built a microgrid system based on photovoltaic power generation, including hydrogen energy storage and battery energy storage, and proposed a microgrid scheduling strategy combining day ahead scheduling and day-to-day real-time scheduling. In the day ahead scheduling stage, in order to improve the timeliness of the microgrid, using the sparrow search algorithm based support vector machine model (SSA-SVM) to predict the day ahead photovoltaic power generation. Taking the minimum use cost of the microgrid as the goal and considering the reliability of the system operation, the multi- objective sparrow search algorithm is used to formulate the day ahead optimal scheduling strategy of the microgrid. In the intra day dispatching stage, the microgrid operation strategy is adjusted in real time according to the actual power generation to eliminate the impact of prediction error. Finally, the feasibility of the prediction algorithm and scheduling strategy is verified by a practical example. The results show that the proposed method can effectively predict the data, reduce the response time of microgrid scheduling, and improve the economy and stability of system operation.