本文旨在改善並聯型主動式電力濾波器(Shunt Active Power Filter, SAPF)之直流鏈電壓(DC-Link)控制策略，以智慧型控制方法改善傳統控制使用上的限制。 由於電感性元件與非線性負載於現今社會中廣泛應用，容易導致系統出現電流諧波、三相不平衡與功率因數偏低等問題，並聯型主動式電力濾波器，能有效作為改善電力品質不佳的解決方案。 傳統比例積分控制在直流鏈電壓控制中，容易出現動態響應不佳與穩定性不足等問題。為此，本文提出結合希爾伯特黃轉換特徵(Hilbert-Huang Transform, HHT)與模糊神經網路控制策略(Fuzzy Neural Network, FNN)之智慧型控制策略，並推導其神經網路架構、線上學習法則與收斂性分析。 本文於MATLAB & Simulink R2021b實現三相四線電路、非線性不平衡負載與變流器控制，模擬正常與失真市電的情境，並比較不同鎖相迴路與控制策略之性能表現，最終透過OPAL-RT硬體迴圈完成即時模擬驗證。實驗結果顯示，使用雙二階廣義積分鎖相迴路鎖相迴路(Double Second Order Generalized Integrator PLL, DSOGI-PLL)時，於正常或失真市電條件下，所提出之智慧型控制策略，能有效降低電流總諧波失真、電流三相不平衡與功率因數校正，並減少直流鏈電壓之震盪響應，以驗證控制策略有效性。;This paper proposes an improved DC-link voltage control strategy for Shunt Active Power Filters (SAPFs) and ameliorates the limitations of traditional control approaches by incorporating intelligent control methods. Due to widespread use of inductive components and nonlinear loads in modern society, issues such as current harmonics, three-phase imbalance, and low power factor often arise in power systems. The shunt active power filter (SAPF) serves as an effective solution to improve poor power quality. Traditional PI controllers often exhibit poor dynamic response and limited stability. To overcome these drawbacks, an intelligent control strategy combining Hilbert-Huang Transform (HHT) and Fuzzy Neural Network (FNN) is developed, including the derivation of its network structure, online learning rules, and convergence analysis. The proposed method is implemented in MATLAB & Simulink R2021b to simulate both normal and distorted grid conditions. Performance under various phase-locked loop (PLL) schemes is evaluated. Real-time validation is conducted using the OPAL-RT hardware-in-the-loop platform. Results confirm that the proposed control method with Double Second Order Generalized Integrator (DSOGI-PLL) significantly reduces total harmonic distortion, improves current unbalanced rate, power factor, and suppresses DC-link voltage oscillations.
