臺灣博碩士論文加值系統

這篇論文首先將光復校區所有智慧電表紀錄到的累計用電度數數據取出並進行數據預處理，包含數據合併、數據轉換及數據清理，將數據內部有問題的數據用空值取代。接著會將經過預處理後的光復校區北區及南區數據進行異常數據檢測及數據填補。最後用LSTM模型預測北區及南區未來30天的用電量，很多能源管理方法都需要以未來用電量為基準，所以選擇用電量預測作為應用。用LSTM模型預測未來30天用電量，會比較輸入1筆到最多2年的數據時的預測結果，以及分成三種輸入參數不同的模型，分別是單純用電量預測模型、代表人員活動的學期參數模型及代表天氣狀況的天氣參數模型。預測出來後再以圖像化及RMSE值來呈現。在不同輸入設定下各個模型通常只要有約7筆數據就有不錯的預測結果；學期參數模型則以分成學期間平日、假日及寒暑假四種時間區間預測結果最佳；天氣參數模型中單純用溼度預測結果最佳。用以訓練的數據集只要有一年的數據量就足夠進行預測，更多數據對預測結果影響不大。

關鍵字：用電量數據、數據預處理、LSTM、用電量預測、用電量圖像化

The accumulated electricity consumption data recorded by all smart meters in Guangfu Campus is taken out and preprocessed first, including data merging, data conversion and data cleaning, and replacing the problematic data with null values. Then, the preprocessed data of the northern and southern areas of Guangfu campus will perform abnormal data detection and data filling. Finally, the LSTM model will be used to predict the electricity consumption of the northern and southern areas of Guangfu campus in the next 30 days. Many energy management methods will require future electricity consumption as a benchmark, so the electricity consumption forecast is chosen as an application.Using the LSTM model to predict the electricity consumption in the next 30 days, it will compare the prediction results when inputting data from 1 to 2 years at most, and divide it into three models with different input parameters, namely a simple electricity consumption forecast model, a semester parameter model representing personnel activities, and a weather parameter model representing weather conditions. The prediction results will be presented in the form of images and RMSE values. Under different input settings, each model usually only uses about 7 pieces of data to have good enough prediction results. The best prediction results of semester parameter model are to divide into the four time intervals of weekdays of the semester, holidays of the semester, winter and summer vacations. The best prediction results of weather parameter model are based on the humidity parameters. The data set used for training only needs a year of data to enough to make predictions, and more data has little effect on the prediction results.

keywords：Electricity Consumption Data, Data Preprocessing, LSTM, Electricity Consumption Prediction, Electricity Consumption Visualization

摘要 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
1 緒論 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 研究動機與背景 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 研究目的與預期結果 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 研究範圍與限制 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 研究架構與流程 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 文獻回顧 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1 能源管理與智慧電表 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 能源管理系統 SCADA . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.2 智慧電表 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 資料庫的知識搜尋 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.1 數據預處理 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 遺失數據處理 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.3 評比 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 用電量預測 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 LSTM 模型介紹 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 小結 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3 用電概述與數據預處理 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1 用電基本介紹 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.1 電力能源政策 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.1.2 用電設備管理 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 能源管理系統 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.1 系統內的智慧電表 . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.2 數據取出及合併 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.3 數據預處理 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 小結 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4 用電量數據分析及預測 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1 用電量數據分析 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.1 數據圖像化 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1.2 數據填補及異常數據檢測 . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 用電量預測 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2.1 預測結果與分析 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.2.2 不同的測試設定 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.3 小結 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5 結論與建議 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.1 結論 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.2 建議 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

[1] P. Siano, “Demand response and smart grids—a survey,” Renewable and sustainable energy reviews, vol. 30, pp. 461–478, 2014.
[2] A. Daneels and W. Salter, “What is scada?” 1999.
[3] M. Kermani, B. Adelmanesh, E. Shirdare, C. A. Sima, D. L. Carnì, and L. Martirano, “Intelligent energy management based on scada system in a real microgrid for smart building
applications,” Renewable Energy, vol. 171, pp. 1115–1127, 2021.
[4] M. Anderson, “What is scada?” 2019.
[5] 柯佾寬、林睿珅、林揚淳及黃培華, “智慧電表及再生能源之結合應用於智慧電網
之效益,” 電力電子技術, vol. 16, no. 6, pp. 63–71, 2018.
[6] U. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, “The kdd process for extracting useful
knowledge from volumes of data,” Communications of the ACM, vol. 39, no. 11, pp. 27–
34, 1996.
[7] S. Sagiroglu and D. Sinanc, “Big data: A review,” in 2013 international conference on
collaboration technologies and systems (CTS). IEEE, 2013, pp. 42–47.
[8] S. García, S. Ramírez-Gallego, J. Luengo, J. M. Benítez, and F. Herrera, “Big data preprocessing: methods and prospects,” Big Data Analytics, vol. 1, no. 1, pp. 1–22, 2016.
[9] S. García, J. Luengo, and F. Herrera, Introduction. Cham: Springer International Publishing, 2015, pp. 1–17. [Online]. Available: https://doi.org/10.1007/978-3-319-10247-4_1
[10] C. Beeri, P. A. Bernstein, and N. Goodman, “A sophisticate’s introduction to database
normalization theory,” in Readings in artificial intelligence and databases. Elsevier,
1989, pp. 468–479.
[11] S. Patro and K. K. Sahu, “Normalization: A preprocessing stage,” arXiv preprint arXiv:
1503.06462, 2015.
[12] S. Van Buuren, Flexible imputation of missing data. CRC press, 2018.
[13] A. Jadhav, D. Pramod, and K. Ramanathan, “Comparison of performance of data imputation methods for numeric dataset,” Applied Artificial Intelligence, vol. 33, no. 10, pp.
913–933, 2019.
[14] C. J. Willmott, “Some comments on the evaluation of model performance,” Bulletin of the
American Meteorological Society, vol. 63, no. 11, pp. 1309–1313, 1982.
[15] 劉家銘, “LSTM 應用於大專院校館舍用電量預測之研究—以交通大學為例,” 交通大
學土木工程系所學位論文, pp. 1–63, 2020.
[16] 易文豪, “EMD-GRU 應用於校舍用電量之預測—以陽明交通大學為例,” 陽明交通大
學土木工程系所學位論文, pp. 1–55, 2021.
[17] 李宏毅, “什麼是深度學習-機器學習的卷土重來,” 數理人文, no. 10, pp. 20–29, 2016.
[18] S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural computation, vol. 9,
no. 8, pp. 1735–1780, 1997.
[19] J. Brownlee, Deep learning for time series forecasting: predict the future with MLPs, CNNs
and LSTMs in Python. Machine Learning Mastery, 2018.
[20] N. Hebbar, “Time-Series-Forecasting-LSTM,” From the GitHub: https://github.com/
nachi-hebbar/Time-Series-Forecasting-LSTM, 2021.
[21] W.-C. Lin and C.-F. Tsai, “Missing value imputation: a review and analysis of the literature
(2006–2017),” Artificial Intelligence Review, vol. 53, no. 2, pp. 1487–1509, 2020.
[22] V. Chandola, A. Banerjee, and V. Kumar, “Anomaly detection: A survey,” ACM computing
surveys (CSUR), vol. 41, no. 3, pp. 1–58, 2009.