Adaptive Control for Single-Phase Unified Power Quality Conditioner Using Neural Networks

Reliable model and high performance controller are of two main issues in research of single-phase Unified Power Quality Conditioner (UPQC). Single-phase UPQC is a nonlinear multi input multi output coupled system and can't be modeled accurately. Therefore neural network with NARMA-L2 structure is utilized to describe the dynamic progress. Based on the NARMA-L2 model, neural feedback linearization controllers are designed for shunt and series active power filters separately. For the DC-link side, a PI-controller is used to maintain the voltage around the reference value. Parameters perturbation and load change simulation experiments are performed to verify the effectiveness, robustness and reliability of the proposed control strategy.