<p>Enabling wind power plants to dynamically regulate the injected reactive power is considered as one of the most promising strategy for enhancing voltage stability in decarbonized power systems. For this purpose, grid codes often require wind power plants to actively participate in voltage regulation through various operating modes, which typically require controlling the injected reactive power at the point of connection based on external set-points fixed by the transmission system operator. In this context, SCADA systems integrate dedicated tools to dispatch the reactive power generated or absorbed by individual wind turbines in order to meet the grid requirements. Solving this challenging issue requires addressing a trade-off between complexity and accuracy, which consists of computing effective dispatch solutions that minimize specific figures of merit (e.g., power losses within the wind power plant) within time frames suitable for real-time operation. To address this problem, this paper proposes a two-layer computational framework in which a linear quadratic regulator is deployed to compute real-time approximate dispatch solutions, while a rigorous optimal reactive power flow problem is solved to update the regulator parameters when its approximation accuracy is expected to deteriorate. Detailed simulation results obtained from a real wind power plant composed of 19 wind turbines of 4.2-MW are presented and discussed to demonstrate the effectiveness of the proposed framework compared to conventional dispatch methods.</p>