Abstract: Grid-connected photovoltaic system is the main development trend of photovoltaic generation systems. The power output of PV is random and uncontrollable because it is influenced by such factors as sunlight and temperature, large-capacity grid-connected photovoltaic power brings certain difficulties in power grid dispatching. In order to connect large-capacity photovoltaic power with power grid, hybrid energy storage is used to smooth power output for improving the schedulability of grid-connected photovoltaic system. A kind of double-deck energy management method is put forward based on different compensatory period in this paper. Taking the cost of energy storage system and the correlation between actual power supply curve and the power dispatching curve of photovoltaic system as the optimized objectives,an improved technology named Pareto genetic algorithm based on niche is applied to obtain optimized solution. In the end, the capacity of energy storage system is optimally equipped by cases, and the rationality and validity of the proposed energy management strategy is analyzed and verified.