應用類神經網路於自我建構式電力系統穩定器之研究

A Study of Self-constructed Power System Stabilizer Using Artificial Neural Network

江慶曜(Ching-Yao Chiang)

指導教授 ： 曾國雄 ； 高文秀

國立臺北科技大學/機電學院/電機工程系所/碩士(2005年)

摘要

由於近年來用電量日益增加，電力系統結構亦趨複雜，電力系統故障或負載變動所引起系統不穩定，必須有一快速反應的電力系統穩定器來改善電力系統穩定度，如何有效的控制電力系統穩定度是很重要的議題。本文提出的類神經網路電力系統穩定器不需分析高階且複雜的電力系統內部狀態，以及不需依賴既有的控制器或參考模式成為學習目標，本文採用非線性自動遞回移動平均模型架構的類神經網路進行受控系統的識別，僅需量測受控系統輸入及輸出相對應之訊號即可進行辨識，將辨識後的網路參數代入控制器訓練網路以進行控制器網路參數的訓練，當控制器網路誤差函數滿足設定之終止條件，即從控制器訓練網路拆解出控制器網路參數，以完成自我建構式類神經網路電力系統穩定器。
最後本文所設計的穩定器與傳統式穩定器，分別以機械功率瞬間變動、輸電線發生對地短路故障及負載變動等擾動模擬與比較，結果顯示本文所設計的穩定器可以有效提升電力系統穩定度。

關鍵字

電力系統穩定器 ； 類神經網路 ； 非線性自動遞回移動平均模型

並列摘要

The capacity of the power system increases gradually in the recent years. The structure of the power system was also become more complicated. The system was unstable which were caused by faulted power system or variable load. It is very important subject to control power system stability effectively. The proposed of power system stabilizer by artificial neural network which is not need to analyze the high order’s power system and power system’s internal states. The proposed power system stabilizer doesn’t depend on existent controller or referenced model for learning goal. The article adopted the nonlinear auto-regressive moving average model to identify the parameters of the pant by artificial neural network. The identification of the model’s parameters is measured from the opposite input/output signal of the plant. After identifying the parameters, substituted the parameters to a part of the controller’s training network for training the controller’s network. The parameters of the controller’s network were separated from the controller’s training network, when the error function of the simulated program satisfied the setting terminated condition. The article accomplished the self-constructed power system stabilizer by using artificial neural network.
Finally, the proposed power system stabilizer was compared with the conventional power system stabilizer. Both stabilizers simulated in mechanical power altered suddenly, transmission line faulted to ground and load alteration individually. The results of the proposed power system stabilizer can promote stability of the power system effectively.

並列關鍵字

Power System Stabilizer ； Artificial Neural Network ； Nonlinear Auto-regressive Moving Average Model

參考文獻

[2] P. M. Anderson and A. A. Fouad, Power system control and stability, Ames, Iowa State University Press, 1977.

[3] IEEE Std.421.5, IEEE Recommended practice for excitation system models for power system stability studies, 1992.

[5] H. A. Moussa and Y. N. Yu, “Optimal power system stabilization through excitation and/or governor control,” IEEE Transactions on Power Apparatus Systems, PAS- 91, pp.1166-1173, April 1972.

[6] S. Juan, L. H. Herron, and A. Kalam, “A fuzzy logic controller applied to power system stabilizer for a synchronous machine power system,” Proceedings of IEEE Region 10 Conference on Computer, Communication, Control and Power Engineering, pp.346-350, November 1992.

[8] E. Handschin, W. Hoffmann, and F. Reyer, “A new method of excitation control based on fuzzy set theory,” Proceedings of IEEE Transactions on Power Systems, vol. 1, pp.533-539, February 1994