Using AI for Closed-Loop Control of a Buck Converter Application

Conference: PCIM Europe digital days 2020 - International Exhibition and Conference for Power Electronics, Intelligent Motion, Renewable Energy and Energy Management
07/07/2020 - 07/08/2020 at Deutschland

Proceedings: PCIM Europe digital days 2020

Pages: 5Language: englishTyp: PDF

Authors:
Slippens, Marvin; Dick, Christian P. (TH Köln, University of Applied Sciences, Germany)

Abstract:
This paper addresses the training of an artificial intelligence (AI) to control the transistor state of a buck converter the most efficient way. For this purpose, a buck converter simulation script was implemented in python. Afterwards, a reinforcement learning agent was trained to provide the optimal duty cycle for the PWM. The trained agent was then tested in the python environment.