ANN Based Sampling Time Selection Criterion for Model Predictive Controlled Multilevel Converter

Model Predictive Control (MPC) is a widely used technique for controlling power converters, but its performance is highly dependent on the appropriate sampling time selection. In high-frequency applications like the nine-level (9L) multi-winding transformer-based boost multilevel converter, improper sampling time selection can lead to increased total harmonic distortion (THD), higher switching losses, and poor dynamic response. This paper investigates the impact of sampling time on the output waveform quality of the converter. It proposes an approach based on artificial neural networks (ANN) to optimize sampling time selection. A comprehensive comparative study is conducted using MATLAB/SIMULINK to analyze the performance of different sampling times on voltage regulation, THD, and computational efficiency. The results demonstrate that selecting an optimal sampling time significantly enhances waveform quality while reducing switching stress and losses. Furthermore, this study highlights the importance of adaptive sampling time selection in MPC-based power converters and presents a data-driven solution for improved performance in high-frequency applications.