State-space Modelling and Stability Analysis of ANN controller for Grid-connected VSC System

Recently, Artificial Neural Network (ANN)-based controllers have been extensively investigated for grid-connected VSC systems. Despite having several advantageous features, the black-box modelling of the ANN controller makes it less industry attractive. This paper reveals the working mechanism of the ANN current controller and its interaction with the VSC system through the small-signal modelling approach. The ANN controller is trained with the standard PI-controller data before deploying for testing. Afterward, state-space equations of the VSC system and ANN controller are derived to develop the small-signal model. The derived model is verified against the time-domain simulation results obtained from the non-linear MATLAB/Simulink model. The small-signal model is then utilised to study the stability of the system. The system and ANN controller behaviour are also studied by performing the participation factor and parametric sensitivity analysis to the small-signal model.