利用粒子群優法設計與電網併聯之風機有效與無效電力自調式控制器

Abstract

本論文之主要目的為利用粒子群優法設計與電網併聯之風機有效與無效電力自調式控制器。傳統控制器常使用固定增益比例積分控制器，主要根據穩態下某一特定工作點來進行設計，當操作於工作點附近時可以得到良好的響應，一旦系統運轉偏離原本工作點過多，例如:風速變動或是三相短路故障等，可能造成系統不穩定。 論文中將以比例積分器做為研究架構，先藉由發電機之功率控制器控制方塊圖，在一個特定工作點之下設計出一組固定增益控制器參數。為了使風力發電機可以得到更好的動態響應，本文選用粒子群優法設計一個自調式控制器以取代固定增益控制器。透過量測發電機當時之狀態，並使用風力發電機的狀態變數微分方程式搭配尤拉法求出狀態變數之數值解，以達到風力發電機實功率動態響應之預測。接著設計一個目標函數作為判斷控制器參數優劣之依據，決定出最佳或近似最佳的控制器參數，達到即時最佳化功率控制器之參數目的，改善固定增益控制器之問題。 本論文研究對象為彰濱地區之離岸風場，將以MATLAB®/Simulink數學軟體建立雙饋式感應風力發電機併網於無限匯流排之模型。本論文在系統發生三相短路故障時提出卸載策略，降低風力發電機輸出之實功率，使其在故障期間提供更多虛功率至電網，穩定系統端電壓。模擬結果會先比較同樣是固定增益控制器在故障後是否執行卸載策略的差異，接著比較固定增益控制器與自調式控制器之動態響應，另外考慮到風速變動下，自調式控制器與固定增益控制器的動態響應差異，最後總結粒子群優法設計之自調式控制器的有效性。

The main purpose of this thesis is to design a self-tuning controller using particle swarm optimization (PSO) for a doubly fed induction generator (DFIG) connected to a power system. A fixed-gain proportional-integral (PI) controller is usually used in conventional controller which is designed based on a particular operating point in steady state. Good dynamic responses can be achieved by the fixed-gain PI controller when the DFIG is operated near the particular operating point. Once the system is operated too far away from the original operating point, such as the cases of wind speed change or three phase ground fault, the system may become unstable. Proportional-integral (PI) controller will be used as the basic control scheme in this thesis. The control block diagram for the DFIG power controller which is derived based on a particular operating point is employed to reach the parameters of the fixed-gain PI controller. In order to have better dynamic responses for the DFIG, a self-tuning controller using PSO algorithm is designed to replace the fixed-gain PI controller in this thesis. To measure the current states of DFIG, the numerical solutions of the state variables are obtained by using the state variable differential equations of DFIG and the Euler method. Therefore, the DFIG real power dynamic responses can be predicted. Then, a proper objective function is chosen as the performance measure to evaluate different PI controller parameters. Finally, we can get the best or nearest optimal controller parameters to achieve the purpose of optimizing power controller parameters in real-time and to improve the dynamic responses of PI power controller. The system under study is a portion of the offshore wind farms in Changhua area. The MATLAB®/Simulink simulation software is employed to build the grid-connected DFIG model. A deloading strategy is proposed in this thesis when system is subject to a three phase ground fault. The purpose of the deloading strategy is to reduce the real power output of the DFIG and increase the reactive power output of the DFIG in order to improve the terminal voltage during fault. The simulation results for the DFIG with and without the proposed deloading strategy are first given. Then, the DFIG dynamic responses using fixed-gain PI controller and the proposed self-tuning controller are compared. In addition, the DFIG dynamic responses obtained from the fixed-gain PI controller and self-tuning controller are also compared under the case of wind speed variation. Finally, the effectiveness of the self-tuning controller using PSO is summarized.