用於改善微電網頻率之雙饋式感應風力發電機粒子群優法自調式控制器

DFIG Self-Tuning Control for Microgrid Frequency Improvement Using Particle Swarm Optimization

楊智翔(Jhih-Siang Yang)

指導教授 ： 許源浴

國立臺灣大學/電機資訊學院/電機工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201802912

摘要

本論文之主要目的在於一微電網系統中，利用粒子群優法設計雙饋式感應風力發電機轉子側之自調式頻率控制器，應用於微電網與無限匯流排因故解聯，使微電網系統形成孤島運轉時，提供暫態期間頻率調節之功用。
傳統風力發電機之輔助頻率控制器設計，大多採用固定比例增益器，且控制器參數為考慮系統之負載變動量、同步發電機及感應風力發電機某一工作點上所設計，其參數為一固定值。倘若其風速、負載變動量過大，使發電機工作點偏移甚多，將會導致微電網系統在孤島運轉期間，調節頻率之響應能力變差，進而造成微電網系統頻率不穩定或使風力發電機失速而解聯，更嚴重時將造成孤島運轉期間微電網系統全黑。
本論文首先將透過小訊號穩定度分析，研究固定比例增益參數對系統特徵值之影響，並利用參與率及靈敏度探討特徵值對於系統狀態變數及控制器參數之影響，並在大擾動時配合暫態穩定度分析，使輔助頻率控制器參數之設計有所依據。
為了改善固定比例增益器在不同系統工作點下之頻率動態響應，本論文提出自調式頻率控制器，透過粒子群優法來即時調整輔助頻率控制器參數，其演算法求解過程中，利用朗吉庫達法預測求解頻率動態方程式之數值解，且透過評估目標函數決定所調整之控制器參數。
本論文以MATLAB/Simulink軟體進行模擬，建立併聯無限匯流排之微電網系統模型，其微電網系統模型包括：同步發電機、雙饋式感應風力發電機以及負載。由模擬結果顯示，當微電網系統因故障與無限匯流排解聯形成孤島運轉暫態期間，在不同負載變動量、同步發電機參數變動、風速變動及控制參數設計不佳時，本論文所提出之自調式頻率控制器相較於固定比例增益器，能改善其微電網系統頻率動態響應，且感應風機也能保持穩定操作。

關鍵字

微電網系統 ； 孤島運轉 ； 雙饋式感應風力發電機 ； 自調式頻率控制器 ； 特徵值 ； 參與率 ； 靈敏度 ； 粒子群優法 ； 朗吉庫達法

並列摘要

The main purpose of this thesis is to design a doubly fed induction generator (DFIG) self-tuning frequency controller for a microgrid using particle swarm optimization (PSO). During transient period of fault, it is applied to provide the frequency adjustment in islanding mode of microgrid due to disconnection between microgrid and infinite bus.
In conventional DFIG frequency support controller, the gain of the proportional controller remains fixed and it is designed based on a particular operating point of system. If the system is operated too far away from the nominal operating point due to large disturbances, the excursions can result in poor capability of adjusting frequency and cause undesirable situations of unstable frequency or DFIG stalling. In the worst case, microgrid blackout in islanding operating mode may be observed.
Small signal stability analysis of the DFIG with fixed gain controller is first performed using eigenvalue analysis. Participation factor analysis and sensitivity analysis are employed to evaluate the effect of various state variables and controller parameters on the eigenvalues. Transient stability analysis is also conducted to investigate the dynamic frequency responses under major disturbances. Proper gains for the fixed gain frequency controller are determined based on the results from both small signal stability analysis and transient stability analysis.
In order to improve the dynamic frequency response achieved by a fixed gain DFIG frequency controller, a self-tuning frequency controller with the controller gain adapted in real-time using PSO is proposed. An efficient formula based on Runge-Kutta method is derived to predict the numerical solution of frequency dynamic equation which is used to evaluate the objective function in PSO algorithm and to determine the real-time value of the frequency support controller parameter.
The Matlab/Simulink simulation software is used to build the grid-connected microgrid model including a synchronous generator unit, a DFIG, and loads. The simulation results for microgrid in transient period under load variations, synchronous generator parameter variations, wind speed change, and imperfect design of controller gain are presented. It is concluded from the dynamic responses that the proposed self-tuning frequency controller can offer better frequency dynamic response than the fixed gain controller. It is also found that DFIG can remain in stable operation.

並列關鍵字

microgrid ； islanding mode ； doubly fed induction generator ； self-tuning frequency controller ； eigenvalue ； participation factor ； sensitivity ； particle swarm optimization ； Runge-Kutta method

參考文獻

[1] Global Wind Energy Council (GWEC), “GLOBAL WIND STATISTICS 2017,” http://www.gwec.net/

[2] 經濟部能源局。https://www.moeaboe.gov.tw/

[3] 台灣電力公司。http://www.taipower.com.tw/

[4] O. Soares, H. Goncalves, A. Martins, and A. Carvalho, ‘‘Neural networks based power flow control of the doubly fed induction generator,’’ in Proc. Int. Conf. Power Eng., Energy Elect. Drives, pp. 599–604, 2009.

[5] D. Qian, S. Tong, H. Liu, and X. Liu “Load frequency control by neural-network-based integral sliding mode for nonlinear power systems with wind turbines,” Neurocomputing, vol. 173, pp. 875-885, 2016.