本文提出一種新穎的預同步方法應用於下垂控制微電網，以實現併網操作。當微電網處於孤島模式且市電恢復正常時，由於微電網與市電在相位、頻率與電壓上的非同步性，使得孤島與併網模式間的無縫切換成為一項重大挑戰。若微電網與市電之間未完成同步，即進行併聯操作，將可能產生湧浪電流與電壓波動，導致設備不穩定甚至損壞。 因此，為解決此問題，本文提出一種新的預同步方法，實現相位同步、頻率恢復與電壓恢復，進而達成微電網與市電的安全併聯。此外，為加快同步過程，本文首度引入兩組二元機率勒壤得模糊類神經網路（Binarized Probabilistic Legendre Fuzzy Neural Network, BPLFNN）控制器，分別取代傳統用於頻率與電壓恢復之比例積分（PI）控制器與模糊類神經網路（FNN）控制器。 本文亦推導所提出 BPLFNN 之網路架構與基於誤差反向傳播（Backpropagation, BP）之線上學習演算法。最後，透過實驗驗證本研究所提出之 BPLFNN 型預同步控制方法，能有效實現下垂控制微電網之安全併網，並具備良好的穩定性與即時響應性能。 ;In this study, a novel pre-synchronization method is proposed for a droop controlled microgrid to implement grid connection. When the microgrid operates in islanded mode and intends to reconnect to a normally operated power grid, seamless transition between islanded mode and grid-connected mode is a significant challenge due to the non-synchronous phase, frequency and voltage between the microgrid and power grid. If the phase, frequency and voltage of the microgrid are not synchronized with the power grid, inrush currents and voltage fluctuations may lead to system instability and damage equipment during grid connection. Hence, to address this issue, a novel pre-synchronization method to achieve the phase synchronization, frequency and voltage restoration is proposed for the droop controlled microgrid to implement grid connection. Moreover, to rapidly achieve the pr-synchronization, two binarized probabilistic Legendre fuzzy neural network (BPLFNN) controllers are firstly proposed to replace the traditional proportional-integral (PI) and fuzzy neural network (FNN) controllers for frequency and voltage restoration. The network structure and online learning algorithm based on backpropagation (BP) of the proposed BPLFNN are derived. Finally, the effectiveness of the proposed BPLFNN-based pre-synchronization method to achieve grid connection of the droop controlled microgrid is verified by experimentation.
