Machine Learning Emulation of Model Predictive Control for Modular Multilevel Converters

This article proposes a machine learning (ML)-based emulation of model predictive control (MPC) for modular multilevel converters (MMCs). In particular, the artificial neural network model, trained offline by data collected from traditional fast MPC method, is used to MMCs with high accuracy. With this training, majority computational burden transferred online offline. Therefore, proposed ML can replace role MPC. The experimental results show that ML-based has same performance as conventional but significantly computationally efficient structure. finding letter provides ground many other applications complex controllers in power electronic systems.