臺灣博碩士論文加值系統

本論文主要目的在於利用類神經網路來判斷當電力系統發生故障時，藉由保護機制將故障隔離後，系統是否會恢復穩定，若故障隔離後，系統仍發生不穩定時，透過類神經網路計算所需卸除負載量，使其系統恢復穩定。當電力系統故障發生時，藉由保護電驛動作迅速將故障點隔離，但因故障點隔離後，系統仍有可能處於不穩定的工作點上，而系統的轉子角度將無法收斂進而影響到其他匯流排導致系統受影響區域擴大，因此，為避免此情況發生，本論文研究分析並探討故障隔離後，因故障引起對系統之衝擊仍有可能發生後續對系統不穩定之情況，因此，透過類神經網路進行卸除負載的動作使故障發生後之不穩定的系統恢復到穩定的工作點上，確保大系統持續運轉。
本論文使用基於倒傳遞演算法的類神經網路做為訓練模型，將系統的狀態值，如轉子角度、匯流排頻率、匯流排電壓、各發電機的發電量、各負載的負載量，做為類神經網路所需的樣本資料。藉由減少關聯性較弱的特徵作為樣本資料，並分析經過處理的特徵對於判斷準確率的影響，最後計算出判斷需要卸除的負載量使其準確率可以達到理想準確率值86%，經由本論文之分析研究後，若對樣本資料的特徵進行篩選，準確率更可以達到90%。
本論文藉由類神經網路樣本資料的特徵做為輸入資料並進行訓練，導論出類神經網路之準確率可高達90%，可做為電力調度人員對電力事故發生時所面臨之接續系統可能發生不穩定情況的決策動作加以輔助決策判斷，使其系統能確保持續穩定運轉。

The main purpose of this thesis is to use the method of neural network to evaluate grid stability in case of happening the fault. If the system happens the fault, protection schemes will isolate faulty areas quickly to make sure there is no impact to grid stability. When the fault is being isolated, the system maybe will initial next instability situations. The analysis of neural network will assess to load shedding actions and capacity to make sure the grid can work without influence. Protection Relays take the protection schemes role to isolate the fault, but the system is perhaps in unstable situations. The rotor angle will not in convergence and will influence on others busbar for operation to result unstable areas extending. In this thesis, it studies the possible scenarios based on the fault isolations by using neural network mothed for execution of load shedding strategy to maintain the system stability for safety operation.
This thesis uses neural network based on Backpropagation (BPN) as the training model to collect rotor angle, Busbar frequency, Busbar voltage, generators capacity, and load capacity. It processes the features by reducing of weakness associativity as sampling data to realize the process about how to proceed accuracy of getting load shedding capacity. The purpose of neural network method is expected to have 86% accuracy. By further modeling from lots of input sampling data which processes the features for reducing of weakness associativity to reach up to 90% accuracy.
In machine learning whose accuracy can reach 90%, it can help the system operators to have important information about action plan references to support operators to make grid operation with stability in the system fault moment.

摘要 I
Abstract II
目錄 III
圖目錄 V
表目錄 VIII
第一章 緒論 1
1.1 研究背景與動機 1
1.2 文獻回顧 1
1.3 研究方法 3
1.4 研究軟體 3
1.5 分章內容 4
第二章 類神經網路介紹 5
2.1 簡介 5
2.2 神經元 5
2.3 感知器 6
2.4 活化函數 8
2.5 倒傳遞演算法 11
2.6 梯度下降法(Gradient Descent) 15
2.6.1 隨機梯度下降法 17
2.6.2 批量梯度下降法 17
2.6.3 小批量梯度下降法 18
2.7 過度擬合 19
第三章 故障隔離後之策略 20
3.1 簡介 20
3.2 流程敘述 20
3.2.1 樣本資料產生流程 20
3.2.2 穩定度與卸除負載量之判斷流程 22
3.3 暫態穩定度定義 22
3.3.1 發電機擺動方程式 23
3.3.2 穩定度之等面積法則 25
第四章 案例分析 26
4.1 研究系統架構 26
4.2 研究系統案例 28
4.3 案例分析 29
4.3.1 穩定度樣本：穩定案例 29
4.3.2 卸載樣本 37
4.4 準確率 69
第五章 結論 74
5.1 結論 74
5.2 未來研究方向 74
參考文獻 75
附錄A 78

[1]A.K. Onaolapo, K.T. Akindeji, “Application of Artificial Neural Network for Fault Recognition and Classification in Distribution Network,” in Proc. of 2019 Southern African Universities Power Engineering Conference/Robotics and Mechatronics/Pattern Recognition Association of South Africa (SAUPEC/RobMech/PRASA), South Africa, pp. 299-304, 2019.
[2]李清吟，新卸載策略之研究，行政院國家科學委員會專題研究計畫成果報告，2006年。
[3]L. Yan, G. Xueping, and L. Jun, “A novel discretization scheme combined entropy with rough set theory for transient stability assessment based on ANN,” in Proc. of 2004 International Conference on Power System Technology, Singapore, Vol. 1, pp. 119-122, 2004.
[4]S. Jana and A. De, “A novel zone division approach for power system fault detection using ANN-based pattern recognition technique,” Canadian Journal of Electrical and Computer Engineering, Vol. 40, No. 4, pp. 275-283, 2017.
[5]D. Song, B. Wen, X. Yang, Y. Gu, S. Wei, and S. Ma, “A New Method for Processing and Application of Wide Area Measurement Big Data in a Power System,” in Proc. of 2015 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, China, pp. 137-140, 2015.
[6]A. Q. Santos, R. M. Monaro, D. V. Coury, and M. Oleskovicz, “Real-time closed loop system controlled by an Artificial Neural Network for estimation of the optimal load shedding,” in Proc. of 12th IET International Conference on Developments in Power System Protection (DPSP 2014), Denmark, pp. 1-6, 2014.
[7]S. Lin, C. Xie, B. Tang, R. Liu, and A. Pan, “The data mining application in the power quality monitoring data analysis,” in Proc. of 2016 IEEE 11th Conference on Industrial Electronics and Application (ICIEA), China, pp. 338-342, 2016.
[8]X. Yan, Y. D. Zhao, G. Lin, R. Zhang, K. P. Wong, and F. Luo, “Preventive Dynamic Security Control of Power Systems Based on Pattern Discovery Technique,” IEEE Transactions on Power Systems, Vol.27, No.3, pp. 1236-1244, 2012.
[9]賴冠廷，深度學習方法於電力系統跳機故障分析，國立臺灣海洋大學碩士學位論文，2018年。
[10]黃漢汶，應用類神經網路於電力系統卸載策略之研究，國立中山大學碩士學位論文，2003年。
[11]T. Watanabe and H. Iima, “Nonlinear Optimization Method Based on Stochastic Gradient Descent for Fast Convergence,” in Proc. of 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Japan, pp. 4198-4203, 2018.
[12]Q. Chen and K. Folly, “Effect of Input Features on the Performance of the ANN-based Wind Power Forecasting,” in Proc. of 2019 Southern African Universities Power Engineering Conference/Robotics and Mechatronics/Pattern Recognition Association of South Africa (SAUPEC/RobMech/PRASA), South Africa , pp. 673-678, 2019.
[13]Q. Zheng, J. Fang, Z. Hu, and H. Zhang, “Aero-Engine On-Board Model Based on Batch Normalize Deep Neural Network,” 2019 IEEE Access, Vol. 7, pp 54855-54862, 2019. (SCI, EI)
[14]B. Datta and S. Mitra, “Short term Prediction of Evaporation at a Particular Place using Artificial Neural Networks,” in Proc. of 2019 Amity International Conference on Artificial Intelligence (AICAI), United Arab Emirates, pp. 702-708,2019.
[15]P. P. Panduranga, D.H. Rao, and A. G. Deshpande, “Fault Tolerance Analysis of Neural Networks for Pattern Recognition,” in Proc. of 2007 International Conference on Computational Intelligence and Multimedia Applications (ICCIMA), India, pp. 222-226, 2007.
[16]Siemens Power Technology International, PSS/E Program Operation Manual, Version 34, 2018.
[17]張智星，MATLAB程式設計入門篇，碁峯資訊股份有限公司，2010年。
[18]鄒偉、王振波、王燕妮，深度學習：基于MATLAB的設計實例，北京航空航天大學出版社，2018年。
[19]郭英勝、鄭志宏、龔志銘、謝哲光，實用Python程式設計，第二版，碁峰資訊股份有限公司，2018年。
[20]劉立民、吳建華、陳開煇，Python深度學習，博碩文化股份有限公司，2017年。
[21]葉怡成，應用類神經網路，儒林圖書公司，2001年。
[22]I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning, The MIT Press, 2017.
[23]P. K. Olulope, K. A. Folly, S. Chowdhury, and S. P. Chowdhury, “Transient stability assessment using artificial neural network considering fault location,” in Proc. of 2010 1st International Conference on Energy, Power and Control (EPC-IQ), Iraq, pp. 67-72, 2010.
[24]D. M. Eltigani, K. Ramadan, and E. Zakaria, “Implementation of transient stability assessment using artificial neural networks,” in Proc. of 2013 International Conference on Computing, Electrical and Electronics Engineering (ICCEEE), Sudan, pp. 659-662, 2013.
[25]張斐章、張麗秋，類神經網路導論原理與應用，滄海圖書資訊股份有限公司，第二版，2015年。
[26]黃國源，類神經網路 Neural Networks，全華圖書股份有限公司，第四版，2018年。
[27]T. M. L. Assis, “Power System Dynamic Security Assessment through a Neuro-Fuzzy Scheme,” in Proc. of 2009 15th International Conference on Intelligent System Applications to Power Systems, Brazil, pp. 1-6, 2009.
[28]S. Yadav and S. Shukla, “Analysis of k-Fold Cross-Validation over Hold-Out Validation on Colossal Datasets for Quality Classification,” in Proc. of 2016 IEEE 6th International Conference on Advanced Computing (IACC), India, pp. 1-6, 2016.
[29]J. G. Shang, J. H. Zhang, W. W. Zhao, and J. Liu, “ANN based dynamic voltage security assessment for a practical power system,” in Proc. of IPEC 2007. International Power Engineering Conference, Singapore, pp. 1947-1262, 2007.
[30]A. Hoballah and I. Erlich “Transient stability assessment using ANN considering power system topology changes,” in Proc. of 2009 15th International Conference on Intelligent System Applications to Power Systems, Brazil, pp 1-6, 2009
[31]林士煥、沈鎮南、凌拯民、張原彰、黃昆松、黃川桂，電力系統分析Power System Analysis，高立圖書有限公司，修訂一版，2010年。
[32]P. Kundur, Power system stability and control, McGraw-Hill Professional Pub, 1994.
[33]T. Niimura, H. Ko, H. Xu, A. Moshref, and K. Morison, “Machine learning approach to power system dynamic security analysis,” in Proc. of IEEE PES Power Systems Conference and Exposition, USA, pp.1-5, 2004.