簡易檢索 / 詳目顯示

回結果列表
研究生： 張名豪
Zhang, Ming-Hao
論文名稱： 基於強健資料前處理與多模型機器學習之短期負載預測
Short-Term Load Forecasting Based on Robust Data Preprocessing and Multiple Machine Learning
指導教授： 陳建富
Chen, Jiann-Fuh
羅國原
Lo, Kuo-Yuan
學位類別： 碩士
Master
系所名稱： 電機資訊學院 - 電機工程學系
Department of Electrical Engineering
論文出版年： 2026
畢業學年度： 114
語文別： 英文
論文頁數： 79
中文關鍵詞： 短期負載預測 、資料前處理 、異常偵測 、多模型機器學習
外文關鍵詞： short-term load forecasting, data preprocessing, anomaly detection, multiple machine learning models
相關次數： 點閱：3  下載：0 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 隨著智慧電網發展與再生能源大量併網，精準的短期電力負載預測已成為維持電力系統供需平衡與提升運轉效率的重要技術，實務上的智慧電表資料常受到設備異常、通訊中斷或資料傳輸干擾影響，產生缺失值、突波異常與數值停滯等資料品質問題，進而降低預測模型的可靠度。
    為改善上述問題，本研究提出一套整合強健資料前處理、資料重建與多模型機器學習之短期負載預測框架，資料前處理階段建構多層次異常偵測機制，包括基於 Isolation Forest 之離群值偵測、突波異常偵測與數值停滯異常偵測，並比較不同插值與平滑方法，模型建構階段採用直接多輸出預測策略，將未來一天 96 筆十五分鐘負載值作為預測目標，並運用多種機器學習模型進行訓練與評估。
    結果顯示，在大多數住宅案例中，所提出的框架能降低異常數據的影響並提升預測準確度，對於負載較穩定的工業案例，改善幅度相對較小，但仍可透過驗證集選擇合適的資料重建與預測模型組合，整體而言本研究可提升智慧電表資料於短期負載預測之實務可行性，並作為電力調度、能源管理與需求端策略分析之參考。

    Accurate short-term load forecasting is essential for maintaining power system balance and efficient operation as smart grids and renewable energy continue to expand. However, smart meter data may contain missing values, spike anomalies, and stuck-value anomalies caused by communication interruptions or equipment faults, thereby reducing forecasting reliability.
    To address these issues, this study proposes a short-term load forecasting framework integrating data preprocessing, reconstruction, and multiple machine learning models. Isolation Forest, spike detection, and stuck-value detection are used to identify anomalies, while different reconstruction methods are compared. A direct multi-output strategy is then applied to predict the next-day load profile at 15-minute intervals.
    The results show that the proposed framework reduces the impact of abnormal data and improves forecasting accuracy, particularly for residential loads. Although the improvement is smaller for stable industrial loads, suitable reconstruction–model combinations can still be selected using validation data. Overall, the framework supports the practical application of smart meter data in short-term load forecasting, power dispatch, energy management, and demand-side analysis.

    摘要 i Abstract ii Acknowledgement iii Contents iv List of Tables vi List of Figures vii List of Abbreviations ix List of Symbols x Chapter 1 Introduction 1 1.1 Background and Motivation 1 1.2 Thesis Structure 4 Chapter 2 Data Preprocessing and Prediction Models 5 2.1 Classification of Power Load Forecasting by Forecasting Horizon 5 2.2 Data Preprocessing and Anomaly Detection 7 2.3 Methodological Analysis of Data Partitioning for Time Series 10 Chapter 3 Methodology 14 3.1 Data Preprocessing and Anomaly Detection Strategy 14 3.1.1 Outlier Detection Using Isolation Forest 19 3.1.2 Rule-based Spike Anomaly Detection Using Load Difference 20 3.1.3 Stuck-value Anomaly Detection Based on Consecutive Segment Characteristics 23 3.2 Load Prediction Phase 26 3.3 Performance Evaluation Metrics 29 Chapter 4 Experimental Results and Discussion 32 4.1 Dataset Description 32 4.2 Results and Analysis of Residential Load Forecasting: Cases 1 and 2 34 4.3 Results and Analysis of Industrial Load Forecasting: Cases 3 and 4 43 Chapter 5 Conclusions and Future Work 53 5.1 Conclusions 53 5.2 Future Research Directions 55 References 57 Appendix A 62 Case 1—Residential Load Profile 62 Case 2—Residential Load Profile 63 Case 3—Industrial Load Profile 64 Case 4—Industrial Load Profile 65

    [1] International Energy Agency, Electricity Grids and Secure Energy Transitions: Enhancing the Foundations of Resilient, Sustainable and Affordable Power Systems. Paris, France: OECD Publishing, 2023, doi: 10.1787/455dd4fb-en.
    [2] International Energy Agency, Integrating Solar and Wind: Global Experience and Emerging Challenges. Paris, France: IEA, Sep. 18, 2024. [Online]. Available: https://www.iea.org/reports/integrating-solar-and-wind.
    [3] 經濟部, 「臺灣 2050 淨零轉型『電力系統與儲能』關鍵戰略行動計畫（核定本）」, Apr. 2023. [Online]. Available: https://ncsd.ndc.gov.tw/Fore/nsdn/about0/Work4.
    [4] California Independent System Operator, “What the duck curve tells us about managing a green grid,” California ISO, Folsom, CA, USA, 2016. [Online]. Available: https://www.caiso.com/documents/flexibleresourceshelprenewables_fastfacts.pdf.
    [5] G. M. Pitra and K. S. S. Musti, “Impact analysis of duck curve phenomena with renewable energies and storage technologies,” Journal of Engineering Research and Sciences, vol. 1, no. 5, pp. 52–60, May 2022, doi: 10.55708/js0105006.
    [6] T. Hong and S. Fan, “Probabilistic electric load forecasting: A tutorial review,” International Journal of Forecasting, vol. 32, no. 3, pp. 914–938, 2016, doi: 10.1016/j.ijforecast.2015.11.011.
    [7] U.S. Department of Energy, Advanced Metering Infrastructure and Customer Systems: Results from the Smart Grid Investment Grant Program, Sep. 2016. [Online]. Available: https://www.energy.gov/sites/prod/files/2016/12/f34/AMI%20Summary%20Report_09-26-16.pdf.
    [8] S. R. Khuntia, J. L. Rueda, and M. A. M. M. van der Meijden, “Forecasting the load of electrical power systems in mid- and long-term horizons: A review,” IET Generation, Transmission & Distribution, vol. 10, no. 16, pp. 3971–3977, Dec. 2016, doi: 10.1049/iet-gtd.2016.0340.
    [9] S. Tiwari, “Electrical load forecasting methodologies and approaches,” The Eurasia Proceedings of Science, Technology, Engineering and Mathematics, vol. 19, pp. 1–8, Dec. 2022, doi: 10.55549/epstem.1218629.
    [10] G. F. Casagrande, O. H. Ando Junior, M. O. Oliveira, O. E. Perrone, and J. H. Reversat, “Very short-term electric load forecasting considering climate and temporal variable,” International Journal of Automation and Power Engineering, vol. 3, no. 1, pp. 9–13, Jan. 2014, doi: 10.14355/ijape.2014.0301.02.
    [11] T. A. Alghamdi and N. Javaid, “A survey of preprocessing methods used for analysis of big data originated from smart grids,” IEEE Access, vol. 10, pp. 29149–29171, 2022, doi: 10.1109/ACCESS.2022.3157941.
    [12] S. Wu et al., “Integrated energy system based on Isolation Forest and dynamic orbit multivariate load forecasting,” Sustainability, vol. 15, no. 20, Art. no. 15029, Oct. 2023, doi: 10.3390/su152015029.
    [13] Z. A. Khan et al., “Efficient short-term electricity load forecasting for effective energy management,” Sustainable Energy Technologies and Assessments, vol. 53, Art. no. 102337, Oct. 2022, doi: 10.1016/j.seta.2022.102337.
    [14] N. Huyghues-Beaufond, S. Tindemans, P. Falugi, M. Sun, and G. Strbac, “Robust and automatic data cleansing method for short-term load forecasting of distribution feeders,” Applied Energy, vol. 261, Art. no. 114405, Mar. 2020, doi: 10.1016/j.apenergy.2019.114405.
    [15] A. Forootani, M. Rastegar, and A. Sami, “Short-term individual residential load forecasting using an enhanced machine learning-based approach based on a feature engineering framework: A comparative study with deep learning methods,” Electric Power Systems Research, vol. 210, Art. no. 108119, Sep. 2022, doi: 10.1016/j.epsr.2022.108119.
    [16] R. El-Hadad, Y.-F. Tan, and W.-N. Tan, “Anomaly prediction in electricity consumption using a combination of machine learning techniques,” International Journal of Technology, vol. 13, no. 6, pp. 1317–1325, 2022, doi: 10.14716/ijtech.v13i6.5931.
    [17] J. Wang, C. Gu, and K. Liu, “Anomaly electricity detection method based on entropy weight method and isolated forest algorithm,” Frontiers in Energy Research, vol. 10, Art. no. 984473, Aug. 2022, doi: 10.3389/fenrg.2022.984473.
    [18] Y. Himeur, K. Ghanem, A. Alsalemi, F. Bensaali, and A. Amira, “Artificial intelligence based anomaly detection of energy consumption in buildings: A review, current trends and new perspectives,” Applied Energy, vol. 287, Art. no. 116601, Apr. 2021, doi: 10.1016/j.apenergy.2021.116601.
    [19] R. Kohavi, “A study of cross-validation and bootstrap for accuracy estimation and model selection,” in Proceedings of the Fourteenth International Joint Conference on Artificial Intelligence, vol. 2, Montreal, QC, Canada, 1995, pp. 1137–1143. [Online]. Available: https://www.ijcai.org/Proceedings/95-2/Papers/016.pdf.
    [20] S. Arlot and A. Celisse, “A survey of cross-validation procedures for model selection,” Statistics Surveys, vol. 4, pp. 40–79, 2010, doi: 10.1214/09-SS054.
    [21] M. Hasanov, M. Wolter, and E. Glende, “Time series data splitting for short-term load forecasting,” in PESS + PELSS 2022—Power and Energy Student Summit, Kassel, Germany, 2022, pp. 141–146. [Online]. Available: https://www.vde-verlag.de/proceedings-en/566013024.html.
    [22] M. Schnaubelt, “A comparison of machine learning model validation schemes for non-stationary time series data,” FAU Discussion Papers in Economics, no. 11/2019, Friedrich-Alexander University Erlangen-Nürnberg, 2019. [Online]. Available: https://hdl.handle.net/10419/209136.
    [23] C. Bergmeir, R. J. Hyndman, and B. Koo, “A note on the validity of cross-validation for evaluating autoregressive time series prediction,” Computational Statistics & Data Analysis, vol. 120, pp. 70–83, Apr. 2018, doi: 10.1016/j.csda.2017.11.003.
    [24] Z. Wang, Z. Zhu, G. Xiao, B. Bai, and Y. Zhang, “A transformer-based multi-entity load forecasting method for integrated energy systems,” Frontiers in Energy Research, vol. 10, Art. no. 952420, Jul. 2022, doi: 10.3389/fenrg.2022.952420.
    [25] M. S. Zare, M. R. Nikoo, M. Chen, and A. H. Gandomi, “Capturing complex electricity load patterns: A hybrid deep learning approach with proposed external-convolution attention,” Energy Strategy Reviews, vol. 57, Art. no. 101638, Jan. 2025, doi: 10.1016/j.esr.2025.101638.
    [26] F. T. Liu, K. M. Ting, and Z.-H. Zhou, “Isolation forest,” in Proceedings of the 2008 Eighth IEEE International Conference on Data Mining, Pisa, Italy, 2008, pp. 413–422, doi: 10.1109/ICDM.2008.17.
    [27] Y. Regaya, F. Fadli, and A. Amira, “Point-Denoise: Unsupervised outlier detection for 3D point clouds enhancement,” Multimedia Tools and Applications, vol. 80, no. 18, pp. 28161–28177, Jul. 2021, doi: 10.1007/s11042-021-10924-x.
    [28] L. J. Tashman, “Out-of-sample tests of forecasting accuracy: An analysis and review,” International Journal of Forecasting, vol. 16, no. 4, pp. 437–450, Oct.–Dec. 2000, doi: 10.1016/S0169-2070(00)00065-0.
    [29] A. M. N. C. Ribeiro, P. R. X. do Carmo, I. R. Rodrigues, D. Sadok, T. Lynn, and P. T. Endo, “Short-term firm-level energy-consumption forecasting for energy-intensive manufacturing: A comparison of machine learning and deep learning models,” Algorithms, vol. 13, no. 11, Art. no. 274, Oct. 2020, doi: 10.3390/a13110274.
    [30] A. M. N. C. Ribeiro, P. R. X. do Carmo, P. T. Endo, P. Rosati, and T. Lynn, “Short- and very short-term firm-level load forecasting for warehouses: A comparison of machine learning and deep learning models,” Energies, vol. 15, no. 3, Art. no. 750, Jan. 2022, doi: 10.3390/en15030750.
    [31] C. Wang, T. Bäck, H. H. Hoos, M. Baratchi, S. Limmer, and M. Olhofer, “Automated machine learning for short-term electric load forecasting,” in Proceedings of the 2019 IEEE Symposium Series on Computational Intelligence, Xiamen, China, 2019, pp. 314–321, doi: 10.1109/SSCI44817.2019.9002839.
    [32] B. Ibrahim, L. Rabelo, E. Gutierrez-Franco, and N. Clavijo-Buritica, “Machine learning for short-term load forecasting in smart grids,” Energies, vol. 15, no. 21, Art. no. 8079, Oct. 2022, doi: 10.3390/en15218079.
    [33] P. Manandhar, H. Rafiq, E. Rodriguez-Ubinas, and T. Palpanas, “New forecasting metrics evaluated in Prophet, Random Forest, and Long Short-Term Memory models for load forecasting,” Energies, vol. 17, no. 23, Art. no. 6131, Dec. 2024, doi: 10.3390/en17236131.
    [34] P. Koponen, J. Ikäheimo, J. Koskela, C. Brester, and H. Niska, “Assessing and comparing short term load forecasting performance,” Energies, vol. 13, no. 8, Art. no. 2054, Apr. 2020, doi: 10.3390/en13082054.

    QR CODE