The current wireless communication network for power systems struggles to securely support numerous distributed control terminals. Traditional methods assume ideal measurements or communications, which isn’t practical with growing grid coverage and increasing demands. Integrating 5G technology enhances smart grids’ access, service support, and information transmission. Wireless Networked Control Systems (WNCS) are cost-effective and easy to maintain but face challenges like network delays, data packet loss, and vulnerability to attacks, which can destabilize the system. This paper reviews power wireless control models and, a distributed robust control mechanism for power control services in wireless networks is designed based on the MA-DDPG multi-agent reinforcement learning algorithm framework to maintain voltage near the rated voltage while minimizing reactive power. The effectiveness of the proposed power wireless network control system is validated through simulation.