ANN-Based Surrogate PI and MPC Controllers for Grid-Connected VSC System: Small-Signal Analysis and Comparative Evaluation

The vector proportional-integral (PI) controller is a widely used control method for grid-connected voltage source converters (VSCs); however, it has limitations that can negatively impact dynamic performance in a power electronics-dominated system. On the other hand, model predictive control (MPC) has proven advantageous; however, it requires significant computational resources. To find an intermediate solution, researchers are exploring using artificial intelligence (AI) techniques. This article presents the implementation and effectiveness of the artificial neural network (ANN) as an inner current controller. Both the vector PI and MPC schemes are used as supervisory controllers for the VSC, and the obtained dataset is used to train two different ANN structures, such as the feed-forward neural network (FFNN) and the recurrent neural network (RNN). The black-box modeling of the FFNN and RNN VSC controllers is revealed through the small-signal modeling approach. The derived small-signal model is also used to study the system's stability under different operating scenarios. Furthermore, the impedance frequency response is presented to understand and compare the frequency-domain characteristics of the ANN controllers. The time-domain performance of the controllers is evaluated and compared in MATLAB/Simulink and an experimental test bench setup.