臺灣博碩士論文加值系統

根據聯合國在西元2015年所頒布的巴黎協議，美國等120個國家訂定了減少排放溫室氣體的目標，使再生能源的發展備受關注。歐盟將提高風力發電、太陽能發電等其他再生能源的使用比例，使的原本2030年的再生能源佔比目標由27%上升至32%，然而再生能源的使用往往伴隨著一些問題，使再生能源的利用率無法達到預期，容易造成能源的浪費，更因為再生能源的不穩定性為原本的電網系統帶來極大的考驗。常見的再生能源發電方式為太陽能、風力、水力、地熱等種類，其中又以太陽能發電以及風力發電最受國際的歡迎，但是再生能源有著極大的不穩定性，會隨著天候與其他因素改變發電量，進而影響電網的安全性，因此再生能源往往無法成為基載電力，若能進行發電量與用電量的預測，將延長電力調度可決策的時間，讓再生能源的使用更加容易。再生能源的使用需要搭配一套”可信賴的電力預測與調度平台”來管理不同種類且數量極多的發電/儲電/用電裝置，更需要一種傳輸框架將新舊的發電裝置資訊快速整合。本篇論文提出一套”基於機器學習以IEC61850整合異質電廠及其預測性電力調度方法”，利用機器學習並根據演算法挑選出可靠的特徵進行用電量的預測，本論文利用TC57所發表的IEC61850傳輸協定，提出一套具備高可靠度、預測能力、自動化調度、可擴展性的電力管理與決策平台，更可相容異質電網的傳輸網路框架。本篇論文以台灣北部的發電及負載作為實驗場域，根據本文提出的決策平台，可以順利的預測用電量，其用電預測的平均誤差百分比最低可達2.61％，並實現多種不同發電/儲能/用電裝置的資訊連結以及自動化的電力調度，探討儲能站相較於再生能源於電網中的關係，並可藉由儲能站的調度穩定突發的電力需求所造成電力不足的情況，進而實現本論文所提出的” 基於機器學習以IEC61850整合異質電廠及其預測性電力調度方法”。

According to the Paris Agreement promulgated by United Nations (UN) in 2015, the targets of reducing greenhouse gas emissions have been set by 120 countries including the United States, and the development of renewable energy has attracted much attention from public. European Union (EU) has also claimed to increase the proportion of various renewable energy sources such as wind power and solar power from 27% to 32% in 2030. However, the deployment of renewable energy is often accompanied by some problems, so that the utilization of renewable energy often fails to meet expectations as well as causes the waste of energy. Moreover, the variation of renewable energy has brought great challenges to the original grid systems. The common methods of generating renewable energy are solar energy, wind power, hydropower, geothermal, etc. Among them, solar power and wind power are the most popular internationally. However, renewable energy has great instability, and it is subject to change the power generation with weather, environment and other factors, which will affect the safety of the power grid. Thus, renewable energy cannot be the base load power. If the load of power grids could be predictable, the time of power dispatching operation can be extended to make the deployment of renewable energy more feasible. The deployment of renewable energy requires “a trustworthy power forecasting and dispatching platform" to manage different types and large number of power generation/storage/electrical devices, and a transport framework to quickly integrate the status information of legacy and modern power plants is needed. In this paper, we propose a Machine-Learning Approach to Predictive Power Load Dispatch for IEC61850-Connected Heterogenous Power Plants, to solve the dispatching problem of next-generation grids based on machine learning. The proposed framework selects reliable features to predict electricity consumption. The IEC61850 protocol published by TC57 is used in this paper to enable a power management and decision-making platform with high reliability, predictive ability, automatic scheduling and scalability, and is more compatible with the transmission network framework of heterogeneous power grid. The power generation and load data sets in northern Taiwan are used as the experimental field. The numeric results show that the power consumption can be predicted accurately, reduce the average error to 2.61%. The information link of the energy storage/electrical device and the automated power dispatching, explore the relationship between the energy storage station and the renewable energy in the power grid, and the power shortage caused by the stable power demand of the energy storage station. This research contributes to the design and the deployment of heterogenous power grids in future.

中文學位論文考試審定書 ii
摘要 iii
Abstract v
致謝 vii
第一章 緒論 1
1.1 前言 1
1.1.1 環境保護之重要性 1
1.1.2 再生能源市場及產值趨勢 3
1.1.3 能源調度市場產值 6
1.1.4 小結 7
1.2 研究動機與目的 7
1.2.1 再生能源的不穩定性以及異質電網上的電力調度問題 7
1.2.2 現有的負載預測以及能源調度技術 8
1.2.3 現有智慧電網的不足 9
1.2.4 現有智慧計算技術 10
1.2.5 新世代異質的智慧電網與預測性調度技術 11
1.3 論文架構 11
第二章 文獻探討與技術分析 12
2.1 再生能源的調度與管理技術 12
2.1.1 世界各國的電能管理框架 12
2.1.2 常見的電能能源調度方式 13
2.2 能源資通訊技術介紹 13
2.2.1 IEC61850通訊協定 14
2.2.2 DNP3.0通訊協定 19
2.2.3 小結 20
2.3 類神經網路介紹及運作原理 21
2.4 能源特徵的選取及預測 24
2.4.1 特徵選擇之重要性 24
2.4.2 傳統的預測演算法 25
2.4.3 小結 27
2.5 常見的缺陷數據處理方法 28
第三章 新世代異質智慧電網技術 30
3.1 系統情境 30
3.2 系統軟硬體架構 33
3.3 電網負載及環境資料蒐集分析 37
3.4 電網負載及環境資料前級處理方法 38
3.5 負載特徵選取方法 38
3.6 新世代的異質智慧電網負載預測模型 41
3.6.1 負載預測模型製作 41
3.6.2 負載預測模型訓練 42
3.7 基於IEC61850網路的異質電廠及其預測性電力調度方法 45
3.7.1 IEC61850數據映射方法 45
3.7.2 IEC61850的調度方法 46
3.8 異質網路下的預測性電力調度方法成效分析 52
3.9 小結 54
第四章 實驗測試與分析 55
4.1 實驗目的 55
4.2 實驗設計 55
4.3 異質智慧電網資料交換測試 56
4.3.1 資料交換花費時間 56
4.3.2 協定轉換效能分析 59
4.4 各項特徵選擇方法結果比較與探討 60
4.5 智慧電網負載預測結果 65
4.5.1 數據缺陷修補結果 65
4.5.2 受預測數值消失預測結果 68
4.5.3 數據預測後IEC61850調度結果 69
4.6 實驗數據分析 71
4.6.1 短期負載預測結果 71
4.6.2 中期負載預測結果 74
4.6.3 IEC61850電力調度模擬結果 77
第五章 結論與未來展望 81
5.1 研究結論 81
5.2 未來展望 82
參考文獻 83
個人著作與榮譽 88

[1]維基百科，世界人口，[Online]. Available：https://reurl.cc/rW6vr，2019
[2]新浪科技，末日時鐘只剩兩分鐘 1953年人類面臨最高危險，[Online]. Available：http://tech.sina.com.cn/d/s/2018-01-31/doc-ifyqyqni5804332.shtml，2018
[3]環境資訓中心，國際級報告：暖化守住1.5 °C有可能 但須「前所未有的努力」，https://e-info.org.tw/node/214376，2018
[4]Zion Research Analysis 2015, “Global Smart Grid Market Set for Rapid Growth, To Reach Around USD 120.0 Billion by 2020,” Market Research Store, 2015
[5]施宇承，光連雙月刊，1st ed. vol. 97. Place Published: Taipei，2012
[6]Ren21, “Renewables 2018 Global Status Report,” Renewable Energy Policy Network for the 21st Century , Technical Report, 2018
[7]國家發展委員會，經濟小辭典-再生能源，[Online]. Available：https://reurl.cc/bplYo
[8]行政院重要政策，推動風力發電4年計畫—潔淨能源 乘風而起，[Online]. Available：https://reurl.cc/eYdyK，2018
[9]吳進忠，再生能源預測對於電力調度的重要性，2018
[10]環境資訓中心，打破「智慧」之謎 智慧電網、智慧城市厲害在哪裡？，[Online]. Available：https://e-info.org.tw/node/209332，2018
[11]台灣電力公司資訊揭露，台電供電系統簡介，[Online]. Available：https://www.taipower.com.tw/tc/page.aspx?mid=211，2018
[12]Hulisani Matsila and Pitshou Bokoro, “Load Forecasting Using Statistical Time Series Model in a Medium Voltage Distribution Network,” Proceeding of IECON 2018 - 44th Annual Conference of the IEEE Industrial Electronics Society, pp.4974-4979, 2018
[13]Alper Tokgöz and Gözde Ünal, “A RNN Based Time Series Approach for Forecasting Turkish Electricity Load,” Proceeding of 2018 26th Signal Processing and Communications Applications Conference, pp.1-4, 2018
[14]S.N. Dodamani, V.J. Shetty and R.B. Magadum, “A Short Term Load Forecast Based on Time Series Analysis: A Case Study,” Proceeding of 2015 International Conference on Technological Advancements in Power and Energy, pp.299-303, 2015
[15]Elvisa Bećirović and Marijana Ćosović, “Machine Learning Techniques for Short-Term Load Forecasting,” Proceeding of 2016 4th International Symposium on Environmental Friendly Energies and Applications, pp.1-4, 2016
[16]Cong Feng and Jie Zhang, “Hourly-Similarity Based Solar Forecasting Using Multi-Model Machine Learning Blending,” Proceeding of 2018 IEEE Power & Energy Society General Meeting, pp.1-5, 2018
[17]Mvkola Medykovskvi, Olena Pavliuk and Roman Sydorenko, “Use of Machine Learning Technologys for the Electric Consumption Forecast,” Proceeding of 2018 IEEE 13th International Scientific and Technical Conference on Computer Sciences and Information Technologies, pp.432-435, 2018
[18]Liu Limei and Hou Xuan, “Study of Electricity Load Forecasting Based on Multiple Kernels Learning and Weighted Support Vector Regression Machine,” Proceeding of 2017 29th Chinese Control and Decision Conference, pp.1421-1424, 2017
[19]Mohammad Kaiser Azad, Sihab Uddin and Maen Takruri, “Support Vector Regression Based Electricity Peak Load Forecasting,” Proceeding of 2018 11th International Symposium on Mechatronics and its Applications, pp.1-5, 2018
[20]Xiaokui Sun, Zhiyou Ouyang and Dong Yue, “Short-Term Load Forecasting Based on Multivariate Linear Regression,” Proceeding of 2017 IEEE Conference on Energy Internet and Energy System Integration, pp.1-5, 2017
[21]李清吟、陳大業，“變電所監控裝置即時告警之研究”，中華民國第三十屆電力工程研討會論文集，台灣桃園，2009
[22]林信成，防災型微電網建置，碩士論文，國立高雄應用科技大學電機工程學系，高雄，2017
[23]Gordoncheng's Blog，各國電力調度中心(ISO)、電力交易所(PX)檢索，https://reurl.cc/MEyjp，2010
[24]環境資訊中心，電力調度ABC烈日當頭 台電祭出哪些法寶開源節流？，https://e-info.org.tw/node/213924，2018
[25]柯佾寬，“IEC 61850 通訊協定 GOOSE 功能性研究探討”，電驛協會102年7月會刊，2013
[26]Mark Adamiak, Drew Baigent, and Ralph Mackiewicz, “IEC 61850 Communication Networks and Systems in Substations”, The Protection & Control Journal, pp. 61-68, 2010
[27]Paul Smart，How to Access Your Measurement Data Using DNP3，[Online]. Available：https://www.campbellsci.com.au/blog/access-measurement-data-using-dnp3，2016
[28]Intelligentcomputing，DNP3協議簡單介紹及協議識別方法，[Online]. Available：https://kknews.cc/zh-tw/tech/x56z4ng.html，2017
[29]王瀚、林純鑫，“智慧型電網與變電所自動化簡介”，電力人，第十五期，2014
[30]尹相志，SQL Server 2005 Data Mining資料採礦與Office 2007資料採礦增益集，悅知文化，2007
[31]SJ，JavaScript 實現類神經網路 (瀏覽器 Deep Learning 好棒棒)，[Online]. Available：https://reurl.cc/pkvjd，2017
[32]Haykin, S, “NEURAL NETWORKS a Comprehensive Foundation,” Canada: Ontatio, 1999
[33]S. Alelyani, J. Tang, and H. Liu. Feature Selection for Clustering: A Review. Data Clustering: Algorithms and Applications, Editor: Charu Aggarwal and Chandan Reddy, CRC Press, 2013
[34]Zhao Z, Morstatter F, Sharma S, Alelyani S, Anand A, Liu H, Advancing feature selection research—Asu Feature Selection Repository. Technical Report, School of Computing, Informatics, and Decision Systems Engineering, Arizona State University, 2010
[35]董道廷，電力系統短期負載預測，碩士論文，國立中山大學電機工程學系，高雄，2017
[36]林建甫，迴歸分析，滄海書局，2017
[37]Deep Systems，Evolution: From Vanilla RNN to GRU & LSTMs，[Online]. Available：https://reurl.cc/z96py。
[38]HNFXS，貝茲曲線應用-貝塞爾插值，[Online]. Available：https://reurl.cc/AZ063，2013
[39]lsxpu，Cubic spline-三次樣條插值，[Online]. Available：https://blog.csdn.net/lsxpu/article/details/38849775，2014
[40]維基百科，XPath，[Online]. Available：https://zh.wikipedia.org/wiki/XPath，2019
[41]楊凱、侯豔、李康，隨機森林變數重要性評分及其研究進展，[Online]. Available：http://www.paper.edu.cn/releasepaper/content/201507-212，2015
[42]Jiapeng Yin, Jian Hou, Zhiyong She, Chengming Yang and Han Yu, “Improving The Performance of SVM-RFE on Classification of Pancreatic Cancer Data,” Proceeding of 2016 IEEE International Conference on Industrial Technology, pp.956-961, 2016
[43]IT獨白者，遞迴式特徵消除：Recursive feature elimination(RFE)，[Online]. Available：https://blog.csdn.net/sun_wangdong/article/details/80863932，2018
[44]台灣電力公司，各種發電方式之發電成本，[Online]. Available： https://www.taipower.com.tw/tc/page.aspx?mid=196，2019
[45]International Energy Agency, “The Power of Transformation”, Technical Report, 2014
[46]能源資訊平台，儲能設施是否可以負擔?，[Online]. Available：http://eip.iner.gov.tw/msn.aspx?datatype=YW5hbHlzaXM=&id=MTYy，2018
[47]華人百科，RMSE，[Online]. Available：https://www.itsfun.com.tw/rmse/wiki-7752717-6710596，2019
[48]BOB L. WALL and JEFF K. ELSER, “Imputation of Missing Data For Input To Support Vector Machines”, [Online]. Available：https://reurl.cc/7aGL9
[49]維基百科，龍格現象，[Online]. Available：https://reurl.cc/dZm1y，2019
[50]M H M R Shyamali Dilhani, Chawalit Jeenanunta, “Daily Electric Load Forecasting: Case of Thailand,” Proceeding of 2016 7th International Conference of Information and Communication Technology for Embedded Systems, pp.25-29, 2016
[51]Ali Lahouar, Jaleleddine Ben Hadj Slama, “Random Forests Model for One Day Ahead Load Forecasting,” Proceeding of 2015 6th International Renewable Energy Congress, pp.1-6, 2015
[52]行政院，本院新聞-關切2025非核家園前之電力系統 賴揆親自主持會議，[Online]. Available：https://reurl.cc/7xM4D，2017
[53]Zhongsheng Guo, “Daily Variation Law of Solar Radiation Flux Density Incident on the Horizontal Surface” Journal of Earth Science & Climatic Change, Guo, J Earth Sci Clim Change 2017, Vol. 8, No. 9 DOI: 10.4172/2157-7617.1000412, 2017
[54]Solar News，太陽光電發電量計算法，[Online]. Available：https://reurl.cc/Yz040，2011