The inherent multi-time-scale fluctuation of renewable energy is the main reason that restricts its large-scale application. In order to improve the overall economic benefit and energy efficiency of the energy storage system. This paper proposes a hybrid energy storage system with compressed air energy storage, battery and supercapacitor. Firstly, analyzes the characteristics of supercapacitors and batteries, and the cooling and heating effects of compressed air system are also analyzed. Secondly, the energy management strategy of hybrid energy storage system using active energy storage strategy is studied. Using extreme leaning machine (ELM) to predict the distribute energy power generation in long time scale and determine the target power of HESS. EEMD method is used to distribute the target power of compressed air, battery and supercapacitor, then according to the constraints of energy storage frequency range, power range, state of energy and other conditions to select the optimal demarcation frequency. Finally, simulation, comparison and analysis are carried out to verify the effectiveness of the proposed method.