Hello FlexSim Community,
We are working on a FlexSim 2025-based simulation model for a pilot production workshop, and I’m currently exploring the possibility of creating an AI-based optimizer for the production process using the results of the FlexSim simulation and external machine learning resources, such as Stable-Baselines3 and reinforcement learning in Python.
I’m aware that FlexSim itself does not provide a built-in AI optimizer, and I’ve reviewed the documentation about the Reinforcement Learning (RL) Tool introduced in FlexSim 2022. However, I would greatly appreciate any practical advice beyond what's covered in the manual.
My main questions:
Has anyone developed a custom RL/AI-based optimizer using FlexSim simulation data?
Are there any working examples or best practices for integrating FlexSim with Python-based RL libraries (e.g., PPO from Stable-Baselines3)?
Any tips on designing observation/action/reward structures within the FlexSim RL Tool?
Any known challenges or lessons learned from building such hybrid simulation-ML systems?
This is our first AI-integrated simulation, and I’d love to hear your experience, feedback, or suggestions on how to move forward effectively.
Thanks in advance for your support!
Best regards,
Oleksii Kovalenko
