臺灣博碩士論文加值系統

在非侵入式負載偵測的研究中，我們要從單一節點的聚合電量資料中辨識出每一個電器，會需要擷取各個電器的特殊特徵，並以單一特徵與單一辨識演算法來進行電器識別，而在以往的研究中，通常會採取高頻取樣的方式來擷取這些特殊的特徵，如高頻的電流諧波或電磁干擾，利用這些特徵來使辨識演算法的準確率提高。然而，利用高頻取樣的方式來擷取特徵進行電器識別需要負擔昂貴的硬體成本，對於一般家庭來說，龐大的機器費用會大幅的降低他們使用此技術的意願，使得非侵入式負載偵測的技術難以普及。
在本論文中，我們使用取樣頻率低且價格相對較便宜的設備來取得負載信號，並從這些資料中擷取出低頻率的特徵，針對可加性的低頻取樣特徵來進行電器識別，而在辨識演算法的部分我們採用Adaboost集成學習模型，將多個低頻特徵配合多個辨識演算法所得到的辨識結果形成一個委員會，藉由投票的機制來取得初步的投票結果，並將此結果當作訓練資料，透過Adaboost演算法來訓練出Adaboost集成學習模型，經由此模型，最終的辨識結果為多個特徵與多個不同權重的辨識演算法來共同決定。
最後我們經由實驗結果發現，Adaboost集成學習模型可以解決典型集成學習模型在相同票數或是相近票數的問題，其辨識準確率也比任何單一辨識演算法還要來的好，透過Adaboost集成學習模型可以使最終的辨識模型得到更好的辨識效果。

In this paper, a non-intrusive appliance load monitoring (NILM) scheme based on the Adaboot ensemble algorithm for cheaper and low frequency meter is developed. In order to apply the NILM scheme we need to extract features for appliances. However, it is a challenging task if we want to know the states for each appliance at home just from information of single point aggregate power meter. In literature, it is usually done by applying high frequency meter to extract high frequency feature, e.g., harmonics and electromagnetic interference, to make recognition accuracy better. However, the hardware of high frequency were costly. For typical family, expensive devices would make the NILM impractical and infeasible.
To develop a NILM that can be applied on a cheaper and low frequency meter, low frequency features should be used. In addition, these low frequency features should satisfy the additivity property in order to be used in our learning model. The Adaboost ensemble learning model is then used as the recognition algorithm in our work. Multiple features and multiple recognition algorithms are used to get initial recognition results. These results are used as the training data, and adopted the Adaboost ensemble learning model. In this model, the recognition result was decided from multiple features and multiple different weight recognition algorithms.
Adaboost ensemble learning algorithm could solve the problem of similar number of votes in typical ensemble learning model. Results show that the proposed Adaboost ensemble learning model could enhance the recognition accuracy.

致謝詞 i
摘要 ii
Abstract iii
目錄 iv
圖目錄 vi
表目錄 viii
第一章 緒論 1
1.1研究動機 1
1.2研究目的 3
1.3論文架構 5
第二章 理論背景 6
2.1電器負載偵測 6
2.1.1侵入式負載偵測(Intrusive Load Monitoring) 7
2.1.2非侵入式負載偵測(Non-Intrusive Load Monitoring) 7
2.2NILM相關研究 9
2.2.1數據採集 9
2.2.2特徵擷取 9
2.2.3辨識演算法 13
2.3集成學習 16
2.3.1建構集成學習模型 16
2.3.2學習演算法問題 18
2.3.3集成學習類型 19
第三章 研究方法與架構 24
3.1系統架構概述 24
3.2訓練預測模型 25
3.2.1特徵擷取 26
3.2.2辨識演算法 30
3.3集成學習模型 36
3.3.1Ensemble learning 36
3.3.2Adaboost 38
3.4電器識別 47
3.4.1事件偵測 47
3.4.2變化事件特徵擷取 47
3.4.3電器識別演算法 48
3.4.4辨識準確率 49
第四章 模擬環境與實驗結果 51
4.1實驗數據 51
4.2實驗結果 53
4.2.1單一預測模型 53
4.2.2集成學習預測模型 55
4.2.3類似特徵電器比較 57
第五章 結論與未來展望 58
參考文獻 59

[1]G.W. Hart, “Nonintrusive Appliance Load Monitoring,” Proceedings of the IEEE, Vol. 80, No. 12, pp. 1870-1891, 1992.
[2]M. Zeifman, K. Roth, “Nonintrusive appliance load monitoring: Review and outlook,” IEEE Transactions on Consumer Electronics, Vol. 57, No. 1,pp. 76-84, February 2011.
[3]A. Zoha, A. Gluhak, M.A. Imran, S. Rajasegarar, “Non-Intrusive Load Monitoring Approaches for Disaggregated Energy Sensing: A Survey,”Sensors,16838-16866, December 2012.
[4]S. Darby, “The Effectiveness of Feedback on Energy Consumption: A Review for Defra of the Literature on Metering, Billing and Direct Displays” Technical Report for Environmental Change Institute, University of Oxford,2006.
[5]K. Ehrhardt-Martinez, K.A. Donnelly, J.A. Laitner, “Advanced Metering Initiatives and Residential Feedback Programs: A Meta-Review for Household Electricity-Saving Opportunities,” Technical Report E105 for American Council for an Energy-Efficient Economy (ACEE), 2010.
[6]M. Weiss, A. Helfenstein, F. Mattern and T. Staake, “Leveraging Smart Meter Data to Recognize Home Appliances”, IEEE International Conference on Pervasive Computing and Communications, pp. 190-197, 2012.
[7]A.G. Ruzzelli, C. Nicolas, A. Schoofs and G.M.P. O’Hare, “Real-Time Recognition and Profiling of Appliances through a Single Electricity Sensor”, 7th Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communications and Networks (SECON), pp. 1-9, 2010.
[8]M. Berges, E. Goldman, H.S. Matthews and L. Soibelman, “Training Load Monitoring Algorithms on Highly Sub-Metered Home Electricity Consumption Data,” 12th International Conference on Computing in Civil and Building Engineering, pp. 406-411,October 2008.
[9]A. I. Albicki, and A. Cole, “Data Extraction for Effective Non-Intrusive Identification of Residential Power Loads,” IEEE Instrumentation and Measurement Technology Conference, pp. 812-815, 1998.
[10]A. Albicki, and A. Cole, “Algorithm for Non-Intrusive Identification of Residential Appliances,” Circuits and Systems, Proceedings of the IEEE International Symposium on, Vol. III, pp. 338-341, 1998.
[11]H.Y. Lam, G.S.K. Fung, W.K. Lee, “A Novel method to construct taxonomy electrical appliances based on load signatures,” IEEE Trans. Consumer. Electron. , Vol 53, pp. 653–660, 2007.
[12]C. Laughman, K. Lee, R. Cox, S. Shaw, S. Leeb, L. Norford and P. Armstrong, “Power Signature Analysis”, IEEE Power and Energy Magazine, Vol. 1, No. 2, pp. 56-63, 2003.
[13]S.B. Leeb, S.R. Shaw, J.L. Kirtley, “Transient Event Detection in Spectral Envelope Estimates for Nonintrusive Load Monitoring,” IEEE Transactions on Power Delivery, Vol. 10, No. 3, pp.1200-1210, July 1995.
[14]K.D. Lee, S.B. Leeb, L.K. Norford, P.R. Armstrong, Jack Holloway, and S.R. Shaw, “Estimation of Variable-Speed-Drive Power Consumption from Harmonic Content,” IEEE Transactions on Energy Conversion, Vol. 20, No 3,pp. 566-574, September 2005.
[15]W. Wichakool, A.T. Avestruz, R.W. Cox, S.B. Leeb, “Modeling and Estimating Current Harmonics of Variable Electronic Loads,” IEEE Transactions on Power Electronics, Vol. 24, No12, pp. 2803-2811, December 2009.
[16]S.N. Patel, T. Robertson, J.A. Kientz, M.S. Reynolds, and G.D. Abowd,“At the Flick of a Switch: Detecting and Classifying Unique Electrical Events on the Residential Power Line,” Conference on Ubiquitous Computing, pp. 271-288, 2007.
[17]S. Gupta, M.S. Reynolds, S.N. Patel, “ElectriSense: Single-Point Sensing Using EMI for Electrical Event Detection and Classification in the Home,” Conference on Ubiquitous Computing, pp. 139-148, 2010.
[18]J. Liang, S.K.K. Ng, G. Kendall, J.W.M. Cheng, “Load Signature Study—Part I: Basic Concept, Structure, and Methodology,” IEEE Transactions on Power Delivery, Vol. 25, No. 2, pp. 551-560, 2010.
[19]J. Liang, S.K.K. Ng, G. Kendall, J.W.M. Cheng, “Load Signature Study—Part II: Disaggregation Framework, Simulation, and Applications,” IEEE Transactions on Power Delivery, Vol. 25, No. 2, pp. 561- 569, 2010.
[20]K. Suzuki; S. Inagaki, T. Suzuki, H. Nakamura, K. Ito, “Nonintrusive Appliance Load Monitoring Based on Integer Programming,” In Proceedings of SICE Annual Conference, pp. 2742–2747, August 2008.
[21]J.Z. Kolter, S. Batra, A.Y. Ng, “Energy disaggregation via discriminative sparse coding,” Advances in Neural Information Processing Systems, pp.1153-1161, 2010.
[22]A. Marchiori, D. Hakkarinen, Q. Han, L. Earle, “Circuit-level load monitoring for household energy management,” IEEE Pervasive Computing, Vol. 10, No. 1, pp.40–48,2011.
[23]D. Srinivasan, W. Ng, A. Liew, “Neural-network-based signature recognition for harmonic source identification,” IEEE Transactions on Power Delivery, Vol. 21, No. 1, pp. 398–405, 2006.
[24]A.G. Ruzzelli, C. Nicolas, A. Schoofs and G.M.P. O’Hare, “Real-Time Recognition and Profiling of Appliances through a Single Electricity Sensor”, 7th Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communications and Networks (SECON), pp. 1-9, 2010.
[25]T.G. Dietterich, “Ensemble Methods in Machine Learning,” Lecture Notes in Computer Science, Berlin: Springer, pp. 1-15, 2000.
[26]L. Breiman, “Bagging Predictors,” Machine Learning, Vol. 24, No. 2, pp. 123-140, 1996.
[27]Y. Freund and R.E. Schapire, “A decision-theoretic generalization of on-line learning and an application to boosting,” Journal of Computer and System Sciences, vol. 55, pp. 119–139, 1997.
[28]A.L. Blum and R.L. Rivest, “Training a 3-node neural network is NP-complete,” Lecture Notes in Computer Science, Vol. 661, pp. 9-28,1993.
[29]L. Hyafil and R.L. Rivest, “Constructing optimal binary decision trees is NP-complete,” Information Processing Letters , Vol. 5, No. 1, Pages 15–17, May 1976.
[30]K. Hornik, M. Stinchcombe, H. White “Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks,” Neural Networks , Vol. 3, No. 5, Pages 551–560, 1990.
[31]M. Kearns, “Thoughts on Hypothesis Boosting, ” Machine Learning class project, December 1988
[32]R.E. Schapire, “The Strength of Weak Learnability,” Machine Learning ,Vol. 5, No. 2, pp. 197-227, June 1990.
[33]Nest Thermostat URL:
https://nest.com/thermostat/meet-nest-thermostat/
[34]yokogawa power meter URL:
http://tmi.yokogawa.com/files/uploaded/Catalog_YMI130_EN_1.pdf