臺灣博碩士論文加值系統

隨著溫室效應影響加劇全球暖化的幅度，淨零排放、永續與節能減碳等環保議題受到重視。為降低能源的浪費，企業加強對能源管理系統的研發，以利於瞭解與控管設備的能源使用，其中透過需量預測 有利於對能源應用的規劃。需量預測可使用電戶簽訂合理的契約容量，並防止超出契約容量與確認參與需量反應的可能性，以優化能源管理系統的電力調度。
本論文之需量預測系統，透過智慧電錶收集用電數據，將負載功率轉換成每日最大需量後進行正規化處理，並加入模糊化歷史溫度特徵，導入預測模型中進行訓練，測試不同超參數對預測模型的影響，選出最佳超參數組合。
本論文依據均方根誤差、平均絕對值誤差與平均絕對百分差3種評估指標，比較灰色預測、遞迴神經網路與長短期記憶模型的誤差率，最終提出基於溫度特徵模糊化的長短期記憶預測模型作為日前需量預測系統之預測模型。

With the intensification of the greenhouse effect and the increase of global warming, environmental protection issues such as net-zero emissions, sustainability, energy conservation, and carbon reduction have received attention. To reduce the waste of energy, enterprises have begun to enhance the research on energy management systems to help understand and control the energy use of equipment. Demand forecasting is helpful for energy application planning. Users are able to estimate a reasonable contracted capacity based on demand forecasting to prevent exceeding contractual capacity and confirm the possibility of participating in demand response programs so as to optimize the power dispatching of the energy management system.
The demand forecasting system in this paper collects electricity consumption data through smart meters. The collected load power data is converted into the daily maximum demand. Normalization processing is performed for fuzzyfication of historical temperature, as well as the forecasting model. The effect of hyperparameters on the prediction model, and the best combination of hyperparameters is studied.
Three evaluation indicators: Root Mean Square Error, Mean Absolute Error and Mean Absolute Percentage Error, are used in this thesis. The performances of gray prediction, recurrent neural network and long short-term memory models are compared. The long short-term memory model with the temperature fuzzification as the final model for the demand forecasting system is proposed in this thesis.

摘　要 i
ABSTRACT ii
致謝 iv
目 錄 v
表目錄 vii
圖目錄 viii
第 1 章 緒論 1
1.1 研究動機與目的 1
1.2 論文架構 2
第 2 章 能源管理系統 4
2.1 建築能源管理系統 4
2.2 需量反應 6
2.2.1 需量反應方案 6
2.2.2 自動化需量反應 10
2.3 負載預測 11
2.3.1 統計預測模型 12
2.3.2 機器學習預測模型 14
2.3.2.1 支持向量機 15
2.3.2.2 類神經網路 17
2.3.2.3 遞迴神經網絡 19
2.3.2.4 卷積神經網路 23
第 3 章 系統架構與研究方法 27
3.1 數據蒐集與預處理 28
3.1.1 皮爾森積動差相關係數 28
3.1.2 正規化 29
3.1.3 模糊邏輯 31
3.2 預測模型 34
3.2.1 灰色預測 34
3.2.2 機器學習 36
3.2.2.1 長短期記憶 36
3.2.2.2 超參數 39
3.3 預測模型評估指標 45
第 4 章 系統架構與結果驗證 46
4.1 系統流程 46
4.2 數據蒐集 48
4.3 數據預處理 56
4.4 模型訓練 59
4.4.1 灰色預測 60
4.4.2 機器學習 64
4.4.2.1 超參數選擇 64
4.4.2.2 模型與特徵 70
4.5 預測結果討論 73
第 5 章 結論與未來展望 75
5.1 結論 75
5.2 未來展望 76
參考文獻 77

[1] 國家發展委員會(2022)。臺灣2050淨零排放路徑及策略總說明。檢自https://www.ndc.gov.tw/Content_List.aspx?n=FD76ECBAE77D9811 (May.05, 2022)
[2] 許志義(2013)。我國電力需求面管理之探討。檢自https://www.ghsu.url.tw/wp-content/uploads/2013/04/DSM001.pdf (May.21, 2022)
[3] U.S. Energy Information Administration. (2021). Annual Electric Power Industry Report, Form EIA-861 detailed data files. https://www.eia.gov/electricity/data/eia861/ (May.25, 2022)
[4] International Energy Agency. (2022). Data browser. https://www.iea.org/fuels-and-technologies/electricity (May.25, 2022)
[5] 黃維綱，民105。實習「需量競價交易流程與調度作業之研習」出國報告。臺灣電力公司實習報告。臺灣電力公司。
[6] 台灣電力公司(2018)。需量反應負載管理措施。檢自https://www-ws.gov.taipei/Download.ashx?u=LzAwMS9VcGxvYWQvMzQyL3JlbGZpbGUvMzY5NjcvNzY4MDEyNS9mZDNhMzc4ZS1lZjIwLTRjN2ItYjg2MS0zMTg2MmUwOTZjZmMucGRm&n=6ZyA6YeP5Y%2BN5oeJ6LKg6LyJ566h55CG5o6q5pa957Ch5aCxLnBkZg%3D%3D (April.23, 2022)
[7] 經濟部能源局(2020)。109年度公用售電業節約用電及減碳成果。檢自https://www.moeaboe.gov.tw/ECW/populace/content/wHandMenuFile.ashx?file_id=8822 (May.15, 2022)
[8] 邢博竣，開放式自動化需量反應介面開發與空調機之人體熱舒適度探討，碩士論文，國立臺北科技大學能源與冷凍空調工程研究所，臺北，2016。
[9] 資訊工業策進會(2013)。AMI與需量反應衍生服務效益及挑戰。檢自https://www.slideserve.com/swain/ami (May.26, 2022)
[10] About OpenADR. (2018). from https://www.openadr.org/overview (May 30,2022)
[11] Y. Yao, H. Xu, Y. Chang, S. Wang and G. Zhou, "Study on Ultra-Short Term Power Load Forecasting Based on Local Similar Days and Long Short-Term Memory Networks," 2019 Chinese Automation Congress (CAC), 2019, pp. 322-327, doi: 10.1109/CAC48633.2019.8996639.
[12] Y. Ren, M. Zhou, G. Li and Z. Shen, "Hierarchically coordinated ultra-short term load forecasting for load aggregator," 2015 5th International Conference on Electric Utility Deregulation and Restructuring and Power Technologies (DRPT), 2015, pp. 141-146, doi: 10.1109/DRPT.2015.7432225.
[13] M. Lekshmi and K. N. A. Subramanya, "Short-Term Load Forecasting of 400kV Grid Substation Using R-Tool and Study of Influence of Ambient Temperature on the Forecasted Load," 2019 Second International Conference on Advanced Computational and Communication Paradigms (ICACCP), 2019, pp. 1-5, doi: 10.1109/ICACCP.2019.8883005.
[14] S. Singh, S. Hussain and M. A. Bazaz, "Short term load forecasting using artificial neural network," 2017 Fourth International Conference on Image Information Processing (ICIIP), 2017, pp. 1-5, doi: 10.1109/ICIIP.2017.8313703.
[15] L. Hu, L. Zhang, T. Wang and K. Li, "Short-term load forecasting based on support vector regression considering cooling load in summer," 2020 Chinese Control And Decision Conference (CCDC), 2020, pp. 5495-5498, doi: 10.1109/CCDC49329.2020.9164387.
[16] O. Olabode, O. Amole, T. Ajewole and I. Okakwu, "Medium-Term Load Forecasting In A Nigerian Electricity Distribution Region Using Regression Analysis Techniques," 2020 International Conference in Mathematics, Computer Engineering and Computer Science (ICMCECS), 2020, pp. 1-5, doi: 10.1109/ICMCECS47690.2020.240907.
[17] R. Kumar, R. Ranjan and M. C. Verma, "New Approach for Long Term Electricity Load Forecasting for Uttarakhand State Power Utilities using Artificial Neural Network," 2021 10th International Conference on System Modeling & Advancement in Research Trends (SMART), 2021, pp. 601-607, doi: 10.1109/SMART52563.2021.9675305.
[18] Box, G. E. P., and David A. Pierce. “Distribution of Residual Autocorrelations in Autoregressive-Integrated Moving Average Time Series Models.” Journal of the American Statistical Association, vol. 65, no. 332, 1970, pp. 1509–26. JSTOR, https://doi.org/10.2307/2284333. Accessed 25 Jun. 2022.
[19] X. Jin, Y. Dong, J. Wu and J. Wang, "An Improved Combined Forecasting Method for Electric Power Load Based on Autoregressive Integrated Moving Average Model," 2010 International Conference of Information Science and Management Engineering, 2010, pp. 476-480, doi: 10.1109/ISME.2010.124.
[20] L. Zhang, P. Shi, X. Lai, P. Du, F. Wen and Y. Liu, "Air-conditioning load forecasting based on seasonal decomposition and ARIMA model," 2021 International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE), 2021, pp. 664-669, doi: 10.1109/ICACITE51222.2021.9404621.
[21] Yeh James(2017)。支援向量機(Support Vector Machine)介紹。檢自https://reurl.cc/1ZL92Y(May.02, 2022)
[22] J. L. Viegas, S. M. Vieira, J. M. C. Sousa, R. Melício and V. M. F. Mendes, "Electricity demand profile prediction based on household characteristics," 2015 12th International Conference on the European Energy Market (EEM), 2015, pp. 1-5, doi: 10.1109/EEM.2015.7216746.
[23] S. R. Abbas and M. Arif, "Electric Load Forecasting Using Support Vector Machines Optimized by Genetic Algorithm," 2006 IEEE International Multitopic Conference, 2006, pp. 395-399, doi: 10.1109/INMIC.2006.358199.
[24] Z. Zhang and S. Ye, "Long term load forecasting and recommendations for China based on support vector regression," 2011 International Conference on Information Management, Innovation Management and Industrial Engineering, 2011, pp. 597-602, doi: 10.1109/ICIII.2011.418.
[25] 李元皓，基於暫態電流圖像與卷積神經網路之非侵入式負載辨識系統，碩士論文，國立臺北科技大學自動化科技研究所，臺北，2021。
[26] M. Dai et al., "A neural network short-term load forecasting considering human comfort index and its accumulative effect," 2013 Ninth International Conference on Natural Computation (ICNC), 2013, pp. 262-266, doi: 10.1109/ICNC.2013.6817982.
[27] D. Horii, A. Ito and T. Nose, "Analysis of Feature Extraction by Convolutional Neural Network for Speech Emotion Recognition," 2021 IEEE 10th Global Conference on Consumer Electronics (GCCE), 2021, pp. 425-426, doi: 10.1109/GCCE53005.2021.9621964.
[28] B. Jeevan, E. Naresh, B. P. V. kumar and P. Kambli, "Share Price Prediction using Machine Learning Technique," 2018 3rd International Conference on Circuits, Control, Communication and Computing (I4C), 2018, pp. 1-4, doi: 10.1109/CIMCA.2018.8739647.
[29] D. Salman, M. Kusaf and Y. K. Elmi, "Using Recurrent Neural Network to Forecast Day and Year Ahead Performance of Load Demand: A Case Study of France," 2021 10th International Conference on Power Science and Engineering (ICPSE), 2021, pp. 23-30, doi: 10.1109/ICPSE53473.2021.9656827.
[30] LHBlog (2018)。RNN遞迴神經網路初始與詳解。檢自https://iter01.com/166306.html (May.20, 2022)
[31] H. Zheng, X. Zhao, Y. Wu, X. Song, K. Jin and Y. Li, "Research on Cold Load Forecasting Model Based on Long Short-Term Memory," 2021 2nd International Conference on Computer Science and Management Technology (ICCSMT), 2021, pp. 87-91, doi: 10.1109/ICCSMT54525.2021.00025.
[32] W. Zheng, Q. Wan, L. Zhu, N. Lv and M. Li, "Short Term Power Load Forecasting Based on Clustering and Long Short Term Memory Network," 2021 IEEE 4th International Electrical and Energy Conference (CIEEC), 2021, pp. 1-6, doi: 10.1109/CIEEC50170.2021.9510632.
[33] S. Yang, X. Yu and Y. Zhou, "LSTM and GRU Neural Network Performance Comparison Study: Taking Yelp Review Dataset as an Example," 2020 International Workshop on Electronic Communication and Artificial Intelligence (IWECAI), 2020, pp. 98-101, doi: 10.1109/IWECAI50956.2020.00027.
[34] Q. Wang, R. Lin, Y. Zhao and H. Zou, "Electricity Consumption Forecast Based on Empirical Mode Decomposition and Gated Recurrent Unit Hybrid Model," 2020 IEEE 20th International Conference on Communication Technology (ICCT), 2020, pp. 1670-1674, doi: 10.1109/ICCT50939.2020.9295955.
[35] I. Atik, "A New CNN-Based Method for Short-Term Forecasting of Electrical Energy Consumption in the Covid-19 Period: The Case of Turkey," in IEEE Access, vol. 10, pp. 22586-22598, 2022, doi: 10.1109/ACCESS.2022.3154044.
[36] A. Agga, A. Abbou, M. Labbadi and Y. el Houm, "Short-Term Load Forecasting: Based on Hybrid CNN-LSTM Neural Network," 2021 6th International Conference on Power and Renewable Energy (ICPRE), 2021, pp. 886-891, doi: 10.1109/ICPRE52634.2021.9635488.
[37] D. H. Vu, K. M. Muttaqi and A. P. Agalgaonkar, "Assessing the influence of climatic variables on electricity demand," 2014 IEEE PES General Meeting | Conference & Exposition, 2014, pp. 1-5, doi: 10.1109/PESGM.2014.6939377.
[38] S. Hochreiter and J. Schmidhuber, "Long Short-Term Memory," in Neural Computation, vol. 9, no. 8, pp. 1735-1780, 15 Nov. 1997, doi: 10.1162/neco.1997.9.8.1735.
[39] 交通部中央氣象局(2022)。觀測資料查詢。檢自https://e-service.cwb.gov.tw/HistoryDataQuery/ (June.06, 2022)
[40] 交通部中央氣象局(2022)。觀測資料數據。檢自https://e-service.cwb.gov.tw/HistoryDataQuery/MonthDataController.do?command=viewMain&station=466920&stname=%25E8%2587%25BA%25E5%258C%2597&datepicker=2022-02&altitude=5.3m (June.06, 2022)
[41] 氣象資料開放平台(2022)。一般天氣預報-今明36小時天氣預報。檢自https://opendata.cwb.gov.tw/dataset/forecast/F-C0032-001 (May.12, 2022)
[42] B. Alaskar et al., "Next-day Electricity Demand Forecast: A New Ensemble Recommendation System Using Peak and Valley," 2021 IEEE Power & Energy Society Innovative Smart Grid Technologies Conference (ISGT), 2021, pp. 1-5, doi: 10.1109/ISGT49243.2021.9372158.