本計畫之目標為發展一單級三相式併網型太陽光電系統及二維遞迴式模糊小腦模 型類神經網路智慧型控制之實虛功控制法，於配電系統三相故障期間使用，其中配電三 相故障為相對地故障或相對相故障。太陽能光電系統利用單級三相電流控制電壓源之變 流器完成最大功率點追蹤法以及低電壓穿透能力之功能，其中最大功率追蹤法採用一新 型之增量電導法。配電系統三相故障時，控制器以符合E.ON 低電壓穿透規範的虛功電 流比例補償，調整注入市電端之虛功量，使太陽光電系統產生之實功與注入市電之實功 維持平衡以及保持穩定，並限制變流器輸出電流之最大值。首先於第一年度進行太陽光 電系統的發電特性的探討，發展單級三相電流控制電壓源之變流器、市電角度偵測策略 以及一新型之最大功率追蹤方法，並建立一實驗系統。本計畫第二年就電網發生故障的 類型進行分析，以及發展故障偵測以及相位鎖定的控制等方法，且利用三相電壓大小之 最小值來判斷故障電壓之大小，並且提出一種以正序電壓成分之大小判斷故障電壓大小 之方法與發展低壓穿越之規範，以確保電力系統於故障期間的穩定性。另一方面，單級 三相式併網型太陽光電系統是一具有不確定性的非線性控制系統，且系統操作點可能受 到太陽照度與溫度的變化而變動，因此要獲得明確的系統模型是困難的。因此本計畫第 三年度將設計二維遞迴式模糊小腦模型類神經網路智慧型控制器以控制故障期間的實 虛功，亦達到具低壓穿越能力的實虛功補償控制之要求。本計畫將詳細推導智慧型控制 器架構中各層之函數以及藉由最高陡降法所推導之線上學習法則，並且使用李亞普諾夫 函數證明其收斂性。最後利用實驗成果來驗證所提出智慧型控制器應用於單級三相式併 網型太陽光電系統之實虛功控制在配電系統三相故障時之成效，並與比例積分控制器來做比較。 ;The purpose of this project is to develop a new active and reactive power control scheme using two-dimensional recurrent fuzzy cerebellar model articulation neural network (2D-RFCMANN) intelligent controller for a single-stage three-phase grid-connected photovoltaic (PV) system during grid faults. The presented PV system utilizes a single-stage three-phase current-controlled voltage-source inverter to achieve the maximum power point tracking (MPPT) control of the PV panel with the function of low voltage ride through (LVRT). Moreover, a new incremental conductance (IC) method is developed for the MPPT control. Furthermore, the constraint of the active and reactive power command of the control scheme is according to the ratio of the reactive current in order to meet the LVRT regulations. In the first year of this project, the single-stage three-phase grid-connected PV system, the MPPT method of PV system, the grid synchronization techniques and the experimental system are developed. In the second year of this project, the analysis of grid fault types, the detective method of the grid faults, the design of phase-locked loop (PLL), the analysis of active power and reactive power and the development of LVRT technology during grid faults are carried out. On the other hand, the single-stage three-phase grid-connected PV system can be regarded as a nonlinear system with uncertainties. There are also some other problems of the three-phase grid-connected PV system, such as the change of temperature and irradiance alters system operating point. Thus, it is very difficult to develop a physical nonlinear model for the system. Therefore, in the last yeart, a new active and reactive power control scheme using the 2D-RFCMANN for a single-stage three-phase grid-connected PV system during grid faults with LVRT function is developed. The recurrent network is embedded in the first layer of the 2D-RFCMANN and a Gaussian basis function is used to model the hypercube structure. The online learning laws of 2D-RFCMANN are derived according to gradient descent method. Additionally, specific learning-rate coefficients for network parameters to assure the convergence of the tracking error are derived using Lyapunov function.
