Moving Average-Based Artificial Neural Network Controller for Voltage Source Converter Transient Suppression

Renewable energy sources are extensively incorporated into the conventional electric grid via Voltage Source Converters. Due to the sporadic nature of renewable sources, voltage fluctuations are developed at the input terminals of converters. However, these converters experience switching transients and inrush current issues during voltage fluctuations, degrading converter performance. With this, the electric grid experiences accelerated component aging, increased power loss, and frequent tripping. To mitigate these issues, a novel "Moving Average-based Artificial Neural Network (MA-ANN)" controller that fuses the Moving Average filter’s smoothing feature with Artificial Neural Network adaptive learning capability has been suggested in this study for accurate current regulation in the system. The Input signal with transients is processed using the Moving Average Filter, where the Artificial Neural Network is trained to predict the appropriate switching to regulate the grid current. Analytical results are validated through the proposed controller, which is then simulated and tested in MATLAB/Simulink, where the DC source is modelled as a distributed energy resource (with its DC link connected to the grid through a voltage source converter). The proposed Moving Average-based Artificial Neural Network (MA-ANN) controller outperformed conventional Artificial Neural Network and Proportional–Integral (PI) controllers in suppressing switching transients, minimizing surge inrush current while maintaining accurate current tracking. This controller offers enhanced robustness, reduced converter stress, and improved power quality, making it a promising control strategy for next-generation renewable energy systems with high dynamic variability.