High penetration of distributed generation involved into the power system will result in some deteriorative influences such as poor power quality. An active power compensation model for grid-connected distributed generation system is proposed. This model is based on a hybrid energy storage system（HESS） to compensate the power fluctuation caused by distributed generation. HESS takes advantage of different energy storage technologies to provide satisfying compensation capability with high energy density. A novel dynamic energy optimization algorithm based on energy prediction（DEOEP） is also proposed. By this algorithm， the efficiency of HESS is improved evidently. To obtain compensation power control value tracked by HESS， a fuzzy control strategy is employed. The HESS approach and DEOEP strategy are verified by simulation. The results of simulation show that the proposed model allows higher overall performance with less energy required for HESS and lower cost to be used. The proposed method provides an alternative strategy to enhance existing energy storage devices due to the practicality of this model.