Adaptive control of the virtual synchronous generator by deep neural networks for a wind high power conversion chain

The virtual synchronous generator (VSG) is commonly used to reproduce 
the inertial response of conventional synchronous machines. However, the 
VSG control architecture relies on controller chains, benchmark 
transformations, and parameter settings, including virtual inertia and 
damping, which limit its flexibility in highly dynamic environments. This 
paper proposes an innovative end-to-end control approach based on a neural 
network to fully replace the classical VSG control structure. The neural 
network developed is trained to directly generate inverter control signals 
from real-time electrical measurements, including voltages and currents, as 
well as active and reactive power. A dataset is generated from a detailed 
VSG model under different operating conditions, and then a multilayer 
neural network is trained using supervised learning with MATLAB. The 
resulting model is then integrated into a complete wind energy conversion 
chain simulated in Simulink. The simulation results demonstrate that control 
based on artificial neural networks ensures better frequency and voltage 
stability, more accurate tracking of the active power injected, and a 
significant improvement in power quality, with total harmonic distortion 
(THD) reduced to 0.04%, compared to 0.51% for conventional VSG control. 
These results confirm the potential of artificial intelligence-based approaches 
for the intelligent control of renewable energy systems.