Energy collaborative optimization of power routing based on PPO and generative adversarial imitation learning

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Renewable Energy Systems

Background:

• The increasing integration of renewable energy sources presents challenges for power system stability and energy management.
• Traditional optimization methods struggle with the intermittent and uncertain nature of distributed energy resources.
• Power routers are crucial for efficient energy utilization and stable power system operation.

Purpose of the Study:

• To develop an advanced energy management strategy for power routers.
• To address the difficulties in managing energy from intermittent renewable sources.
• To enhance the efficiency and stability of power systems with distributed energy.

Main Methods:

• Integration of Proximal Policy Optimization (PPO) with a multi-agent framework.
• Application of Generative Adversarial Imitation Learning (GAIL) with a double-buffer mechanism.
• Optimization of communication and collaboration among multiple agents for energy management.

Main Results:

• The enhanced algorithm achieved stable average round rewards and strategy loss function convergence.
• The model maintained DC bus voltage within a narrow fluctuation range (728V-732V) in practical scenarios.
• The proposed model demonstrated lower electricity costs and total runtime compared to existing methods.

Conclusions:

• The developed algorithm significantly improves energy collaborative optimization for power routers.
• This approach offers a practical solution for energy management challenges in power systems with high renewable energy penetration.
• The study advances the field of intelligent energy management for modern power grids.