|
本論文提出使用類神經網路，進行變壓器磁滯飽和特性模擬的一種新方法
。從二種不同額定下所得的電壓、電流瞬時值，透過監督式學習的過程，
可以獲得精確的變壓器磁滯迴路模型。本論文並將已建立之鐵心特性模型
與有限差分法相結合，進一步模擬暫態湧入電流的特性；分別探討投入角
度、剩磁量、二次側負載等因素，對湧入電流大小及特性所產生的影響。
由於所需之電氣資料是容易獲得的，不像其他方法需要許多複雜的參數，
故較具實際上的適用性；此外與實驗值比較的結果，證實本論文所提出的
磁滯飽和模型，具有良好的準確性及應用性。
This thesis presents a new approach for modeling the magnetic
saturation and hysteresis of transformers using an artificial
neural network. From two sets of field test data, i.e., the
instantaneous voltage versus current curves under different
loading conditions, the hysteresis loops can be approximated to
any desired degree of accuracy through a supervised training
session. The salient feature of this approach is that it is
more general and simpler to implement than the established
known approaches. The effectiveness of the new model is
verified by experimental results. Further- more, incorporating
the proposed model with finite difference method, the transient
inrush currents in transformers can be accurately predicted.
|