The distributed energy storage device units (ESUs) in a DC energy storage power station (ESS) suffer the problems of overcharged and undercharged with uncertain initial state of charge (SOC), which may reduce the service period of ESUs. To address this problem, a distributed secondary control based on diffusion strategy is proposed. In the first layer, each ESUs operates with its local controller by droop control. In the second layer controller, diffusion strategy coordinate the SOC of multiple distributed ESUs with uncertain initial SOC. Compared with centralized control, the proposed method contributes to a higher scalability and reliability. Diffusion strategy allows adjacent nodes to diffuse and cooperate information in real-time, and it includes a stochastic gradient term. Thus, diffusion strategy can achieve a higher convergence rate and lower mean-square-error than consensus strategies. For the second layer controller, first the coordinated control of ESUs is transformed into two optimization problems: the balanced output voltage and the balanced SOC of each ESU. Then, diffusion strategy is used to solve the optimization problems. Finally, a simulation model is built to verify the effectiveness of diffusion strategy. The results show that diffusion strategy always ensure the stability irrespective of the combination topology. Compared with the widely used consensus method, diffusion strategy shows a good performance in realizing demand response for distributed ESUs.