摘 要
近年因大數據、各式分析方法及人工智慧技術的快速發展，如何將此相關的技術應用在電能量測準確度的領域並驗證其可行性，對電能檢測領域來說是全新的技術應用，在人工智慧技術中，又以類神經網路因具有並行處理輸入與輸出數據資料集合的能力及自我學習的優點，而被廣泛且有效應用在各領域的預測分析驗證上。
本研究係探討以不同的瓦時計準確度量測方法，透過倒傳遞類神經網路技術進行準確度誤差差異之研究，以驗證應用類神經網路技術於瓦時量測方法之可行性。瓦時計準確度誤差量測方法係依據國際驗證標準(IEC62053-11、IEC62053-22)所規範之量測方法進行試驗，試驗影響的參數包括準確度、瓦特值、相位角、常數等數據，實驗數據經正規化處理後，代入類神經網路模型工具以預測瓦時計於多相式平衡負載測試方法下之準確度誤差值。
實驗結果顯示，以倒傳遞類神經網路最佳化模型實驗結果之均方誤差值(MSE)為1.1915 × 10-4、均方根差(RMSE)為 0.011、R2為0.96。就本文所探討之倒傳遞類神經網路技術結合瓦時準確度誤差差異之驗證研究而言是可行，且準確度高。

關鍵詞：倒傳遞類神經網路；量測驗證；瓦時計

Abstract
In recent years, due to the rapid development of big data, various analytical methods and artificial intelligence technology, how to apply this related technology in the field of electro-energy measurement accuracy and verify its feasibility, is a new technology application in the field of electric energy testing. In artificial intelligence technology, the Artificial Neural Network(ANN) is widely and effectively applied in various fields of predictive analysis and verification because of its ability to process the collection of input and output data in parallel and the advantages of self-learning.
This research is to approach the method of accuracy measurement with different Watt-hour meters, and the research of accuracy error difference via Back Propagation Neural Network(BPNN), in order to verify the feasibility of applying Artificial Neural Network technology in watt-hour measurement method. The measurement method of the accuracy error of the watt-hour meter is tested according to the measure method of the International Verification Standard (IEC62053-11, IEC62053-22), and the parameters of the test influence include the accuracy, watt value, phase angle, constant etc. After normalization of the experimental data, using the Artificial Neural Network Model Tool to predict the accuracy error value of the Watt-hour meters under test method of the polyphase balanced load.
The experimental results show that the Mean Square Error (MSE) is 1.1915 x 10-4, the Root Mean Square Error (RMSE) is 0.011 and the R2 is 0.96 those which are the optimized model experimental results using the Back Propagation Neural Network. As explored of the difference verification of this study that it is feasible and high accuracy in Back Propagation Neural Network combined with the Watt-hour meter accuracy error.

Keywords：Back Propagation Neural Network；Measurement Validation；Watt-hour meter

參考文獻

[1] 林素琴、林志勳，“我國住宅部門電力使用研究”，臺灣能源期刊第四卷第三期，2017。
[2] 吳再益、林唐裕、侯仁義、柯亮群、黃永慧，“我國能源政策發展模式及其未來方向探討”，臺灣銀行季刊第六十三卷第二期，2012。
[3] 經濟部能源局，2016年能源統計手冊第84頁，2016。
[4] 朱榮貴，“建置智慧型電表基礎建設取代傳統計費電表之關鍵績效指標”，國立臺北科技大學管理學院工業工程與管理EMBA專班，碩士論文，2016。
[5] 李宗恩，“以先進電表基礎建設為基礎之時間電價研究”，國立臺灣科技大學電機工程系，碩士論文，2018。
[6] 丘焙安，“ZigBee應用於智慧電表之開發”，國立臺北科技大學自動化科技研究所，碩士論文，2014。
[7] 葉怡成，「類神經網路應用模式與實作」，儒林圖書有限公司，2002。
[8] 盧苹源，“應用類神經網路技術在金門風場之長時間風力發電預測”，碩士論文，國立臺灣師範大學工業教育學系，2014。
[9] 李育帆，“利用倒傳遞類神經網路預測透水混凝土抗壓強度及透水係數之研究”，碩士論文，逢甲大學土木工程學系，2016。
[10] S. Leva, A. Dolara, F. Grimaccia, M. Mussetta, E. Ogliari, “Analysis and validation of 24 hours ahead neural network forecasting of photovoltaic output power,” IMACS, 2015.
[11] A. Gupta, T. Kim, T. Park, C. Lee, “Intelligent Direct Torque Control of Brushless DC motors for hybrid electric vehicles,” IEEE Vehicle Power and Propulsion Conference, pp. 116-120, 2009.
[12] 國家教育研究院，雙語詞彙、學術名詞暨辭書資訊網，http://terms.naer.edu.tw/
[13] A. Carullo, M. Parvis, A. Vallan, “Measurement of Spurious-related Errors on Commercial Wattmeters,” IEEE Instrumentation and Measurement Technology Conference, pp. 2140-2144, 2001.
[14] IEC62053-11, “Electricity metering equipment(a.c.) – Particular requirements – Electromechanical meters for active energy(class 0.5, 1 and 2),” International Elector technical Commission ,Geneva Switzerland, 2003.
[15] IEC62053-22, “Electricity metering equipment(a.c.) – Particular requirements – Static meters for active energy(class 0.2s and 0.5s),” International Elector technical Commission , 2003.
[16] George A. Smith, Measurement Canada, Paul G. Rivers, Measurement Canada, Volt-Ampere Demand Measurement, 2006.
[17] 林俊良，“微處理式電子式電力電錶的研製”，碩士論文，國立臺灣科技大學工程技術研究所，1994。
[18] 林東山，「電表基本概論」，1976。
[19] 張斐章、張麗秋，「類神經網路導論：原理與應用(第二版)」，滄海圖書，2015。
[20] Ming-Jong Tsai, Chen-Hao Li, “Optimal Laser Cutting Parameters for QFN Packages by Utilizing Artificial Neural Networks and Genetic Algorithm,” Journal of Materials Processing Technology, Vol. 208, pp. 270-283, 2008.
[21] Ming-Jong Tsai, Chen-Hao Li, “Prediction of Laser Cutting Qualities for QFN Strips by Using Levenberg-Marquardt-based Neural Network,” Journal of the Chinese Society of Mechanical Engineers(EI/SCI), Vol. 31, No.4, pp. 273-380, Aug. 2010.
[22] Sanjay Mishra, Vinod Yadava. “Modeling and optimization of laser beam percussion drilling of thin aluminum sheet”, Optics & Laser Technology Volume 4, pp. 461-474, 2013.
[23] 羅華強，「類神經網路-MATLAB的應用第三版」，高立圖書，2013。
[24] 傅心家，「神經網路導論」，第三波文化事業公司，1999。
[25] 張智星，「MATLAB程式設計[入門篇]」，清蔚科技/鈦思科技，2004。
[26] E. B. Lewis, “Control of body segment differentiation in Drosophila by the bithorax gene complex”, Embryonic development, Vol.1, pp. 269-288, 1982.