Intelligent UPQC control using SA-GANN and optimized DC-link regulation in grid-connected microgrids

Abstract This paper presents a photovoltaic (PV)-assisted Unified Power Quality Conditioner (UPQC) for enhancing power quality in grid-connected microgrids. The main novelty lies in the integration of a Sequence-Aware Gated Attention Neural Network (SA-GANN) and a Chaotic Parrot Optimization Algorithm tuned PI (CPOA-PI) controller. Unlike conventional UPQC controllers, the proposed SA-GANN directly learns positive and negative-sequence voltage and current characteristics to generate coordinated reference signals for both series and shunt compensators, eliminating the need for separate sequence extraction and transformation stages. In addition, the CPOA-PI controller provides fast and stable DC-link voltage regulation under dynamic operating conditions. The proposed PV-UPQC system is evaluated in MATLAB/Simulink under no-load, voltage sag, and voltage swell conditions. Simulation results demonstrate effective voltage restoration, harmonic suppression, and DC-link stabilization. The proposed method achieves source current THD values of 1.98%, 2.04%, and 2.50% for the three phases, reduces the voltage unbalance factor to 0.3%, and maintains a power factor of 0.999. The results confirm that the proposed control strategy provides superior power quality enhancement and dynamic performance compared with existing UPQC control approaches.