A Main and Distribution Network Coordinated Voltage and Frequency Control Method Based on Graph Neural Network Agents

By dividing the distribution network into regions and utilizing graph neural network intelligent agent technology, the output power of distributed energy equipment is optimized, solving the problem of insufficient control of the main distribution network power system, realizing precise control of the main grid voltage and frequency, and improving the system's control performance.

CN116071189BActive Publication Date: 2026-05-26THE HONG KONG POLYTECHNIC UNIV SHENZHEN RES INST
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
THE HONG KONG POLYTECHNIC UNIV SHENZHEN RES INST
Filing Date
2022-12-29
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing technologies have poor control over the main and distribution power systems and are unable to effectively cope with the uncertainties of nonlinear systems and complex distributed control behaviors.

Method used

By employing graph neural network agent technology, the power distribution network is divided into several sub-regions, and a target expression is constructed in each region. Using value networks and agent networks, the output power of distributed energy devices is optimized through the control behavior of distributed energy devices to control the voltage and frequency on the main power grid side.

Benefits of technology

It enables precise control of the main and distribution power systems, improving the system's control performance and ability to cope with nonlinear systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116071189B_ABST
    Figure CN116071189B_ABST
Patent Text Reader

Abstract

This invention relates to the field of power grid control technology, specifically to a method for coordinated voltage and frequency control of the main and distribution networks based on graph neural network agents. The invention first divides the distribution network into several sub-regions. Then, for each sub-region, a target expression is constructed based on a value network and an agent network. Next, the network target parameters of each agent device corresponding to the maximum output of the target expression are determined. Finally, the voltage and frequency at the main grid supply point connected to the distribution network are controlled by adjusting the power output of distributed energy devices according to the multiple agent devices. Both the value network and the agent network in this invention are graph neural networks. By independently controlling the power generation of distributed energy devices in each sub-region through multiple agent networks, precise control of the frequency and voltage at the system connection point formed by the main and distribution networks is achieved.
Need to check novelty before this filing date? Find Prior Art