電能是我們生活中不可或缺之能源，其中以冷氣空調耗電量佔整體耗電量之大宗。而為達到全球推動之節約能源政策，安裝智慧電表不僅能監測電器耗電量，亦可利用本文研擬的分析方法預測未來耗電量，提高智慧電表的使用價值。本研究由文獻回顧中得到啟發，以臺灣科技大學營建工程系之研究室為實驗空間，針對建築空間內空調耗能情形進行數據收集，結合環境監測、紅外線感測器和風扇調節系統，探討環境變因對於空調耗電量之影響，並建立耗電量預測模型。由於冷氣空調耗電量資料同時具有線性與非線性的特質，因此本研究提出一混合式模型，結合線性自回歸移動平均(ARIMA)模型以及非線性的萬用啟發式最小平方機器學習(MetaFA-LSSVR)模型推估冷氣空調耗電量，統合稱為自回歸移動平均結合萬用啟發式優化機器學習法(ARIMA- MetaFA-LSSVR)。此法將回歸運算分為兩部分，首先透過ARIMA模型訓練線性預測並計算殘差值，再連同第一部分產出的值和影響耗電量因子套入MetaFA-LSSVR模型中獲得瞬間耗電量預測值。運算過程中，結合滑動視窗法彌補大資料量處理效能的落差，茲以快速預測次日耗電量。而從敏感度分析中得出，整合ARIMA輸出值與考量時間(小時)與紅外線感測頻率屬性時，為此冷氣空調資料庫之最適模型，其每分鐘耗電量預測值與實際值平均單天預測誤差的相關係數(correlation coefficient, R)為62%、方根誤差(root mean square error, RMSE) 為0.41kw、絕對誤差(mean absolute error, MAE) 為0.19kw、總誤差率(total error, TER)為8%。評估結果顯示，本研究提出之自回歸啟發式優化機器學習模型可有效先期預測次日的冷氣空調每分鐘耗電量，茲以作為建物管理單位與使用者於空調設備節電規畫決策資訊，建構智慧建築遠端監測耗能及用電量趨勢預測系統之雛形。

Electricity is an essential form of energy in our lives. Of all the electricity consumed by people, that for air conditioners accounts for the highest proportion. In response to energy conservation policies advocated around the world, we recommend the installation of smart meters. Concerning smart meters, they not only monitor the electricity consumption of electrical appliances, but can also be combined with the analytical method developed in this study to predict future electricity consumption, thereby enhancing the usage value of smart meters. In this study, literature was reviewed to find inspirations and experiments were conducted inside a laboratory of the Department of Civil and Construction Engineering, the National Taiwan University of Science and Technology. Data were collected to identify the air conditioning usage situation in architectural space. Next, an environmental monitoring method, infrared sensors, and fan adjustment systems were used to investigate the effects of environmental variables on the electricity consumption situation of air conditioners as well as to build an electricity consumption forecast model. Because electricity consumption data of air conditioners display both linear and nonlinear characteristics, this study introduced a hybrid model, in which a linear autoregressive integrated moving average (ARIMA) model and a nonlinear nature-inspired metaheuristic-optimized least squares support vector regression (MetaFA–LSSVR) model were combined to estimate the electricity consumption of air conditioners; the hybrid model was called “ARIMA–MetaFA–LSSVR.” This model divided regression operation into two parts, in which the ARIMA model first trained the system to make linear predictions and calculated residual values. Next, values created from the previous step and factors that influenced electricity consumption were incorporated into the MetaFA–LSSVR model to generate forecast values. During the calculation procedure, moving windows were used to make up for poor data processing efficacy and predict electricity consumption for the following day. A sensitivity analysis showed that the optimal ARIMA–MetaFA–LSSVR model (for the air conditioner database used in this study) was obtained by using ARIMA (1,0,2) and considering the time (measured in h) and the infrared sensor frequencies. Concerning the comparison between energy consumption predictions and actual electricity consumptions, the said model was able to achieve a correlation coefficient (R) of 62%, a root mean square error (RMSE) of 0.41, a mean absolute error (MAE) of 0.19, and a total error (TER) of 4%. Therefore, the ARIMA–MetaFA–LSSVR introduced in this study can be used to predict the per-minute electricity consumption of air conditioners for the following day.

[1] A.S. Ahmad, M.Y. Hassan, M.P. Abdullah, H.A. Rahman, F. Hussin, H. Abdullah, R. Saidur, A review on applications of ANN and SVM for building electrical energy consumption forecasting, Renewable and Sustainable Energy Reviews 33 (2014) 102-109.
[2] W. Aslam, M. Soban, F. Akhtar, N.A. Zaffar, Smart meters for industrial energy conservation and efficiency optimization in Pakistan: Scope, technology and applications, Renewable and Sustainable Energy Reviews 44 (0) (2015) 933-943.
[3] D. Bennett, Smart_Grid_Top_Markets_Report 2015, in: I.T. Administration (Ed.), Industry & Analysis, U.S. Department of Commerce, 2015.
[4] S. Bin, H. Dowlatabadi, Consumer lifestyle approach to US energy use and the related CO2 emissions, Energy Policy 33 (2) (2005) 197-208.
[5] D. Chen, S. Kalra, D. Irwin, P.S. Fellow, J. Albrecht, Preventing Occupancy Detection From Smart Meters, IEEE transacitions on smart grid 6 (2015) 208 - 215.
[6] K.-Y. Chen, C.-H. Wang, A hybrid SARIMA and support vector machines in forecasting the production values of the machinery industry in Taiwan, Expert Systems with Applications 32 (1) (2007) 254-264.
[7] J.-S. Chou, N.-T. Ngo, Time series analytics using sliding window metaheuristic optimization-based machine learning system for identifying building energy consumption patterns, Applied Energy 177 (2016) 751-770.
[8] J.-S. Chou, N.-T. Ngo, A.-D. Pham, Shear Strength Prediction in Reinforced Concrete Deep Beams Using Nature-Inspired Metaheuristic Support Vector Regression, Journal of Computing in Civil Engineering 30 (1) (2015) 04015002.
[9] J.-S. Chou, A.-D. Pham, Enhanced artificial intelligence for ensemble approach to predicting high performance concrete compressive strength, Construction and Building Materials 49 (2013) 554-563.
[10] J.-S. Chou, A.-D. Pham, Smart Artificial Firefly Colony Algorithm-Based Support Vector Regression for Enhanced Forecasting in Civil Engineering, Computer-Aided Civil and Infrastructure Engineering 30 (9) (2015) 715-732.
[11] J.-S. Chou, A.S. Telaga, Real-time detection of anomalous power consumption, Renewable and Sustainable Energy Reviews 33 (0) (2014) 400-411.
[12] C.F. Covrig, M. Ardelean, J. Vasiljevska, A. Mengolini, G. Fulli, E. Amoiralis, Smart Grid Projects Outlook 2014, in: C.F. Covrig, M. Ardelean, J. Vasiljevska, A. Mengolini, G. Fulli, E. Amoiralis (Eds.), JRC Science and Policy Reports, European Commission, 2014.
[13] E. Crisostomi, C. Gallicchio, A. Micheli, M. Raugi, M. Tucci, Prediction of the Italian electricity price for smart grid applications, Neurocomputing 170 (2015) 286-295.
[14] S.S.S.R. Depuru, L. Wang, V. Devabhaktuni, A Conceptual Design Using Harmonics to Reduce Pilfering of Electricity, IEEE PES General Meeting (2010) 1-7.
[15] S.S.S.R. Depuru, L. Wang, V. Devabhaktuni, Smart meters for power grid: Challenges, issues, advantages and status, Renewable and Sustainable Energy Reviews 15 (6) (2011) 2736-2742.
[16] M.E. El-hawary, The Smart Grid—State-of-the-art and Future Trends, Electric Power Components and Systems 42 (3-4) (2014) 239-250.
[17] S.M. Farahani, A. Abshouri, B. Nasiri, M. Meybodi, A Gaussian firefly algorithm, INTERNAtional Journal of Machine Learning and Computing 1 (5) (2011) 448.
[18] T.E. Foundation, US-UTILITY-SCALE SMART meter deployments, (2014).
[19] F. Gaxiola, P. Melin, F. Valdez, O. Castillo, Generalized type-2 fuzzy weight adjustment for backpropagation neural networks in time series prediction, Information Sciences 325 (2015) 159-174.
[20] F. Haldi, D. Robinson, The impact of occupants' behaviour on building energy demand, Journal of Building Performance Simulation 4 (4) (2011) 323-338.
[21] B. Halvorsen, B.M. Larsen, The flexibility of household electricity demand over time, Resource and Energy Economics 23 (1) (2001) 1-18.
[22] B.T. Jiang, F.Y. Zhao, Particle swarm optimization-based least squares support vector regression for critical heat flux prediction, Annals of Nuclear Energy 53 (2013) 69-81.
[23] Y. Kaneda, H. Mineno, Sliding window-based support vector regression for predicting micrometeorological data, Expert Systems with Applications 59 (2016) 217-225.
[24] M. Khashei, M. Bijari, A novel hybridization of artificial neural networks and ARIMA models for time series forecasting, Applied Soft Computing 11 (2) (2011) 2664-2675.
[25] M.C. Leung, N.C.F. Tse, L.L. Lai, T.T. Chow, The use of occupancy space electrical power demand in building cooling load prediction, Energy and Buildings 55 (2012) 151-163.
[26] Y. Liao, M. Turner, Y. Du, Development of a Smart Grid Roadmap for Kentucky, Electric Power Components and Systems 42 (3-4) (2014) 267-279.
[27] M.N.Q. Macedo, J.J.M. Galo, L.A.L. de Almeida, A.C. de C. Lima, Demand side management using artificial neural networks in a smart grid environment, Renewable and Sustainable Energy Reviews 41 (2015) 128-133.
[28] S. Marvin, H. Chappells, S. Guy, Pathways of smart metering development shaping environmental innovation, Computers, Environment and Urban Systems 23 (1999) 109-126.
[29] K. Mohammadi, S. Shamshirband, M.H. Anisi, K.A. Alam, D. Petković, Support vector regression based prediction of global solar radiation on a horizontal surface, Energy Conversion and Management 91 (2015) 433-441.
[30] P. Murphy, M. Angemeer, D. Collie, F. N., H. A., D. McFadden, K. Major, N. J., P. Shervill, W. Smith, Enabling tomorrow's electricity system Report, in: I.E.S. Operator (Ed.), Ontario Smart Grid Forum, 2010.
[31] C. Narendra Babu, B. Eswara Reddy, Prediction of selected Indian stock using a partitioning–interpolation based ARIMA–GARCH model, Applied Computing and Informatics 11 (2) (2015) 130-143.
[32] H. Nie, G. Liu, X. Liu, Y. Wang, Hybrid of ARIMA and SVMs for Short-Term Load Forecasting, Energy Procedia 16 (2012) 1455-1460.
[33] D. Ömer Faruk, A hybrid neural network and ARIMA model for water quality time series prediction, Engineering Applications of Artificial Intelligence 23 (4) (2010) 586-594.
[34] S.K. Pal, C.S. Rai, A.P. Singh, Comparative Study of Firefly Algorithm and Particle Swarm Optimization for Noisy Non-Linear Optimization Problems, International Journal of Intelligent Systems and Applications 4 (10) (2012) 50-57.
[35] S. Patnaik, I. Khandelwal, R. Adhikari, G. Verma, International Conference on Computer, Communication and Convergence (ICCC 2015)Time Series Forecasting Using Hybrid ARIMA and ANN Models Based on DWT Decomposition, Procedia Computer Science 48 (2015) 173-179.
[36] A. Saad al-sumaiti, M.H. Ahmed, M.M.A. Salama, Smart Home Activities: A Literature Review, Electric Power Components and Systems 42 (3-4) (2014) 294-305.
[37] S.R. Samantaray, Letter to the Editor: Smart Grid Initiatives in India, Electric Power Components and Systems 42 (3-4) (2014) 262-266.
[38] J. Soares, M.A. Fotouhi Ghazvini, Z. Vale, P.B. de Moura Oliveira, A multi-objective model for the day-ahead energy resource scheduling of a smart grid with high penetration of sensitive loads, Applied Energy 162 (2016) 1074-1088.
[39] J.A.K. Suykens, T.V. Gestel, J.D. Brabanter, B.D. Moor, J. Vandewalle, Least Squares Support Vector Machines, World Scientific Publishing Co Pte Ltd, 2002.
[40] H. Wang, D. Hu, Comparison of SVM and LS-SVM for Regression, 1 (International Conference on Neural Networks and Brain) (2005) 279-283.
[41] X. Wang, M. Han, Improved extreme learning machine for multivariate time series online sequential prediction, Engineering Applications of Artificial Intelligence 40 (2015) 28-36.
[42] G. Wood, M. Newborough, Dynamic energy-consumption indicators for domestic appliances: environment, behaviour and design, Energy and Buildings 35 (8) (2003) 821-841.
[43] Y. X-S, Firefly algorithm.Bristo1, UK:Luniver Press, 2008.
[44] D. Xenias, C.J. Axon, L. Whitmarsh, P.M. Connor, N. Balta-Ozkan, A. Spence, UK smart grid development: An expert assessment of the benefits, pitfalls and functions, Renewable Energy 81 (2015) 89-102.
[45] X.-S. Yang, Analysis of Algorithms, Nature-Inspired Optimization Algorithms, Elsevier, Oxford, 2014, pp. 23-44.
[46] C. Yuan, S. Liu, Z. Fang, Comparison of China's primary energy consumption forecasting by using ARIMA (the autoregressive integrated moving average) model and GM(1,1) model, Energy 100 (2016) 384-390.
[47] W. Zhang, S. Liu, N. Li, H. Xie, X. Li, Development forecast and technology roadmap analysis of renewable energy in buildings in China, Renewable and Sustainable Energy Reviews 49 (2015) 395-402.
[48] K. Zhou, S. Yang, Understanding household energy consumption behavior: The contribution of energy big data analytics, Renewable and Sustainable Energy Reviews 56 (2016) 810-819.
[49] S. Zhou, Z. Wu, J. Li, X.-p. Zhang, Real-time Energy Control Approach for Smart Home Energy Management System, Electric Power Components and Systems 42 (3-4) (2014) 315-326.
[50] 于宗先, 台灣地區住宅與商業部門能源消費調查研究, in: 王京明, 陳澤義, 中華經濟研究院 (Eds.), 能源研究發展基金研究報告, 經濟部能源委員會, 1994.
[51] 台灣行政院, 智慧電網總體規劃方案, in: 經濟部能源局 (Ed.), 經濟部能源局, 臺灣, 2012.
[52] 李天行, 林楓淳, 結合ARIMA 與支援向量迴歸於財務時間序列預測模式之建構－以新加坡交易所日經225 指數期貨為例, 管理學研究所, 天主教輔仁大學, 2007.
[53] 李俊賢, 江泰緯, 混合複數類神經模糊與自動回歸差分平均移動方法之智慧型時間序列預測模型, e-business 15 ( 2013) 137-158.
[54] 張四立, 朱振豪, 住宅部門節電措施及CO2減量之探討-以冷氣空調為例, 自然資源與環境管理研究所, 國立台北大學, 2007.
[55] 畢威寧, 劉亮成, 灰預測在台灣地區電力需求上之應用研究, 科學與工程技術期刊 3 (2) (2007) 11-18.
[56] 陳碧綉, 賈繼德, 台灣電力需求預測模型之探討─ARIMA模型及迴歸模型, 經濟學系, 東吳大學, 2009.
[57] 曾君儒, 省電不難，智慧電表教你玩, 能源報導, 臺灣, 2014.
[58] 賀濤, 周正歐, 基于分形自仿射的混沌时间序列预测, 物理學報 56 (2) (2007) 693-700.
[59] 楊奕農, 時間序列分析：經濟與財務上之應用, 雙葉書廊, 臺灣, 2009.
[60] 福島康裕, 劉怡廷, 台灣冷氣能源消耗減量潛勢於科技、政策及使用者行為不同情境下之探討, 環境工程研究所, 國立成功大學, 2013.