本論文旨在改善微電網在孤島模式時因負載變化而造成的電壓擾動與併網模式時的實虛功控制暫態響應，因此使用一具有模糊推理機制和線上學習能力的模糊類神經網路(Fuzzy Neural Network, FNN)來取代傳統比例積分(Proportional Integral, PI)控制器，該網路經常被應用在處理具有非線性和不確定性的控制系統。 本文模擬採用Solore公司所開發之電路模擬軟體PSIM (Power Simulation)來模擬一微電網之架構且分別操作於孤島模式和併網模式來驗證此演算法之可行性。在實作部分，硬體將以三相三線橋式轉換器作為逆變器之實體架構，將演算法寫入德州儀器(Texas Instruments, TI)之DSP TMS320F28335微控制器，最終驗證所提演算法之可靠性。 ;This paper aims to improve the voltage disturbance caused by the load change at islanding mode and the transient response of the real and reactive power control at grid-connected mode in the microgrid. Therefore, the method proposed in this paper is to replace the traditional proportional integral(PI) controller with a fuzzy neural network(FNN) that has fuzzy inference mechanism and online learning capability. The network is often used to deal with nonlinear and uncertain control systems. In this paper, the circuit simulation software PSIM (Power Simulation) developed by Solore is used to simulate the structure of the microgrid and operate at islanding mode and grid-connected mode respectively to verify the feasibility of this algorithm. For the practical implementation part, the hardware is based on three-phase three-wire bridge converter as the physical structure of the converter, and write the algorithm into the DSP TMS320F28335 micro-controller produced by Texas Instruments (TI) to verify the reliability of algorithm.
