This is an example of using the Q-Agent from the MATLAB Reinforcement Learning Toolbox with a discrete-event-based SimEvents model. It shows a procedure that makes it possible to use the agent in an event-driven manner. It can also be transferred to other agents in the toolbox.
The Simulink model is structured using the concept of Experimental Frame. Please refer to the following publications:
-
Pawletta, T., Bartelt, J. (2023). Integration of Reinforcement Learning and Discrete Event Simulation Using the Concept of Experimental Frame. Eurosim Congress 2023, Amsterdam, Netherlands, July 3-5, PostConf pub. In: SNE Simulation Notes Europe, 33(3):101-109, DOI: 10.11128/sne.33.tn.10651. (full-text on ResearchGate)
-
Pawletta, T., Bartelt, J. (2023). Integrating Reinforcement Learning and Discrete Event Simulation Using the Concept of Experimental Frame: A Case Study with MATLAB/SimEvents. ASIM STS/GMMS & EDU Workshop'23, Magdeburg, Germany, March 6-7, 133-140. DOI 10.11128/arep.21, ASIM Mitteilung 185. (full-text on ResearchGate)