利用類神經網路設計雙饋式感應風力發電機之輔助頻率控制器

Abstract

本論文主要目的在於設計雙饋式感應風力發電機之類神經網路輔助頻率控制器，以改善區域電網於孤島穩態運轉時遭受到一大型負載變動量之頻率響應特性，使系統頻率符合本論文之法規標準與限制條件，也避免觸碰到低頻電驛，造成當地用戶停電。 首先推導區域電網系統頻率控制非線性數學模型，並加入考慮風機線上機組數量，以表示當風機機組故障或檢修時，其慣性將會改變，以及對於區域電網系統之供電能力與頻率響應造成的影響。 接著利用非線性數學模型尋找類神經網路輔助頻率控制器之訓練樣本資料，將其訓練後並且應用於風機之輔助頻率控制器，並且與原先固定不變之固定增益'K' _'PD' 比較頻率響應，驗證類神經網路輔助頻率控制器可以因應區域電網系統參數改變。 本論文藉由MATLAB®/Simulink軟體進行模擬，並以台灣彰化海濱區域電網之非線性數學模型為例，驗證所提出之類神經網路輔助頻率控制器的有效性。

In order to achieve better dynamic frequency response of a local power system, a supplementary frequency controller for a wind farm with doubly fed induction generator (DFIG) is designed using artificial neural network (ANN) in this thesis. Under-freuqnecy load shedding caused by a sudden power unbalance in the local power system can be avoided by the proposed supplementary freuqnecy controller. First, a nonlinear model for the local power system which takes the number of wind generator on-line, wind speed, and load disturbance into account is derived. The dynamic frequency response of local power system under different number of wind generators, different wind speeds, and different load disturbances are analyzed. Next, the training patterns for the ANN are created using dynamic simulations-based on the nonlinear model for the local power system. The connection weights for the ANN are trained using these training patterns. In order to validate the proposed ANN based supplementary frequency controller, the frequency responses from fixed-gain controller are compared with those from the proposed ANN based supplementary frequency controller under different number of wind generators, wind speeds, and load disturbances. In order to demonstrate the effectiveness of the fixed-gain frequency controller and the proposed ANN based supplementary frequency controller, digital simulations using MATLAB®/Simulink are performed on a local power system in Changhua, Taiwan.