Robust Deadbeat Control Scheme for a Hybrid APF With Resetting Filter and ADALINE-Based Harmonic Estimation Algorithm

A novel hybrid active power filter (HAPF) topology is proposed, which shows advantages of the conventional HAPF and the
$LCL$
filter in terms of reduced dc-link voltage, lower switching ripples, and less electromagnetic interference injection. To enhance the noise rejection capability of the digital controller, the resetting filters (RFs) are utilized as prefilters before the analog/digital sampling stage. A robust deadbeat current control law is derived based on the average current tracking scheme, where the effect of the RFs is favorably incorporated. To alleviate the difficulties in current controller design, the ac-side capacitor voltage of the HAPF is estimated by using the adaptive linear neural network identifier in the load current feedforward loop. Hence, a simple proportional controller is utilized in the inner current loop. Moreover, the grid current feedback and load current feedforward strategies are used to achieve precise current tracking and fast dynamic response. A 75-kVA prototype system is built for verification. The validity of the proposed HAPF and its control algorithms are confirmed by the experimental results.