Hybrid Prediction-Based Deadbeat Control for a High-Performance Shunt Active Power Filter

Abstract

This paper proposes a hybrid current predictive deadbeat tracking control method to address the problem that a conventional deadbeat current tracking control has evolved into a deadbeat control in the digital implementation process. First, the reference current prediction link provides in advance the harmonic compensation reference current at two sampling periods by adopting adaptive forward linear prediction and repetitive prediction algorithms. It realizes the switching between the two current prediction algorithms by setting load dynamic occurrence discriminant conditions. Second, the inverter output current historical and harmonic compensation reference current data is used to predict the output current of the inverter. It can effectively predict the output current of the inverter while considering measurement noise suppression. The problems of the deadbeat control method are its strong mathematics model dependence and weak ability to resist high-frequency interference. Therefore, this study analyzes the robust stability of the control system under the uncertainties of the output filter inductor and equivalent resistance parameters of the active power filter and the measurement noise transmission in the system. Finally, the simulation and experimental results indicate that the proposed method based on hybrid current prediction has a high steady-state compensation accuracy and a fast dynamic response speed under steady-state and load mutation conditions. Thus, the proposed method can still effectively compensate the harmonic current under the condition of uncertain output filter inductor and its equivalent resistance parameters.