Unsupervised Learning-Based Inner Control Loop Design for Uncertain Grid-Tied VSC Model

Voltage Source Converters (VSCs) are essential for grid-connected systems, enabling efficient active and reactive power control. However, classic fixed-gain controllers struggle with nonlinear dynamics, parameter uncertainties, and harmonic distortions, leading to performance degradation under non-ideal conditions. This paper introduces an unsupervised learning-based inner control loop (UL-ICL) that dynamically adjusts controller gains and decoupling parameters using fuzzy c-means clustering. By addressing these challenges, the UL-ICL ensures accurate current tracking and robust decoupling, even under dynamic conditions. Simulation results validate its effectiveness, outperforming fixed-gain controllers in precision and resilience. The proposed method offers a promising solution for real-time control of grid-connected VSC systems.