Auto-Tuning of Adaptive Control Initial Gains Using Virtual System-Based Approach in Grid-Tied Converters

Abstract

This paper proposes a novel virtual system-based online tuning strategy for initial gains of adaptive controllers applied to grid-tied converters with LCL filters. The method relies on the implementation of a robust model reference adaptive control (RMRAC) law combined with a full adaptive super-twisting sliding mode action and a disturbance rejection mechanism. To auto-tune the related adaptive gains, a virtual system is excited by a frequency-rich reference signal, ensuring persistent excitation of the regressor and fast convergence of the adaptive gains before the inverter is connected to the physical grid. Once convergence is achieved, the tuned controller is seamlessly transferred to the real plant, where the reference is provided by a grid synchronization unit based on a Kalman filter phase-locked loop. Experimental results demonstrate smooth synchronization, bounded control signals, reduced transient responses, and improved robustness against parametric uncertainties, load disturbances, and grid harmonics, while maintaining acceptable current THD levels. Beyond this case study, the proposed auto-tuning approach can be extended to other RMRAC-based adaptive control schemes, as long as the plant can be properly modeled and simulated, enabling initial gain adjustment without empirical tuning or offline optimization.