Reinforcement Learning Methods

Currently there are no reinforcement learning methods implemented for forecast-based controllers. Feel free to expand the framework with your own reinforcement learning forecast-based controllers!

Potential approaches for reinforcement learning forecast-based controllers could include:

• Deep reinforcement learning with forecast information as part of the state

• Recurrent neural networks that process forecast sequences

• Model-based reinforcement learning that incorporates forecast models

• Multi-agent reinforcement learning for coordinating multiple energy resources based on forecasts

If you implement a reinforcement learning forecast-based controller, consider contributing it back to the HAMLET project.