RL example and questions

[akinar1990]

Issue Type

Select one:

[] Bug report
[X] Question
[] Feature Request
[] Other

Fill these out:

• python version: 3.8.20
• python-microgrid version: 1.4.1

---

Question

• Question:

Hi,

I am trying to train a RL model to control my microgrid.
This section is not finished in the documentation and I wanted to ask if you have example code that I could align to?

Furthermore, I have managed to train a simple A2C model with the library stable-baselines3. For this to work I had to upgrade all to code to use gymnasium and change some outputs from the step and reset functions to align with gymnasium requirements.
However, I don't really understand the purpose of the control algorithm and the priority lists. The general behaviour seems to be implemented in the microgrid.py step() function . Does this essentially mean that the RL algorithm is only selecting which Microgrid controllable to use first via the chosen PriorityList?

How do the Forecasting functions come into play? Can they be used for training an RL algorithm?

Thank you very much for your work. Using python microgrid makes my life a lot easier.

Akin