Abstract: During the 14th Five-Year Plan period, photovoltaic and other clean energy will be in the countryside for large-scale development, but its volatility is strong, poor controllability, and not configured energy storage, not as a reliable power supply in electricity. Therefore, PV combined with flexibility, stability, high energy storage is the trend of the times. Firstly, This paper with the Shanxi Xian yang rainbow of photovoltaic power station history data as an example, based on the sparrow optimization model under different weather forecast the four seasons of photovoltaic power generation, then select the length of memory neural network in the region load forecasts on different time scales, Finally, to maximize the PV given that vary according to the difference of the photovoltaic power generation and load demand, put forward four kinds of energy storage configuration scheme, the results show that the energy storage can improve the PV given and the power supply reliability.