Abstract: In order to solve the problem of considerable delay and error in response to automatic generation control(AGC) instructions in thermal power generating units, a method for optimizing the capacity configuration of a hybrid energy storage system based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN) is developed. First, the power required by the hybrid energy storage system is obtainedby following the AGC instructionand its response curve of the thermal power generating units over a certain period of time. Based on this, the demand power is decomposed by using CEEMDAN to yield the error between the AGC instruction and its response curve of the thermal power generating units at different frequencies. By selecting the appropriate energy storage components, a hybrid energy storage system structure model that cooperates in the response of the AGC instructions in the thermal power plant is constructed, and the power distribution is carried outunder the condition of considering the working characteristics of two different types of energy storage equipment: energy-type lithium iron phosphate battery and power-type flywheel battery. Finally, a capacity optimization configuration model is established with the objective of minimizing overall costs while accounting for constraints such as the state of charge(SOC) of the energy storage system, capacity, and charge/discharge power. The results of the power decomposition are optimized synergistically with the capacity configuration model to determine the optimal power allocation and corresponding energy storage configuration scheme. An example is demonstrated to verify the effectiveness of the proposed method in compensating for the delayed power response resulting in thermal power generating units with high reliability and economic efficiency. And compared with a single energy storage element, the hybrid energy storage system designed in this scheme has better economy.