Abstract: This paper addresses the voltage control issue in 35 kV distribution networks by proposing an intelligent voltage regulation scheme based on hierarchical control. The scheme integrates reinforcement learning with adaptive control, where the upper layer employs deep reinforcement learning for coordinated control of multiple transformers, and the lower layer uses adaptive control to ensure precise adjustment of individual transformers. Simulation results demonstrate that the scheme exhibits excellent performance in terms of voltage stability and disturbance rejection, enhancing the operational quality and efficiency of the distribution network.