從用電安全、英國智慧電表開放資料、數位分身的角度開始，本研究探討台灣智慧電表的加值應用。過去已和能源局與工研院合作取得少量台灣資料來分析用電行為，本研究改良深度學習LSTM方法，使其先預測用電不安全的三種情形：瞬間用電超標、持續性用電超標，與經常性用電超標，建立預警系統，接著，結合IoT裝置收集環境面資料，搭配智慧電表來進行用電量預測。最後，在測試的場域建構虛擬世界，以數位分身的作法示範節電規則，讓居民可在個人隱私保障的情況下，預測未來是否用電安全，與可進行的節電行為為何。 ;Starting from the perspective of electricity safety, UK smart meter open data, and digital avatar, this study explores the value-added application of Taiwan's smart meters. In the past, we have cooperated with the Energy Bureau and ITRI to obtain a small amount of Taiwan smart meter data to analyze the behavior of electricity. This study improves the deep learning LSTM method, so that it can predict the three situations of unsafe electricity: instantaneous power consumption exceeds the standard, and continuous and frequent power consumption exceeds the standard. Additionally, we will establish an early warning system for unsafe electricity usage, and then combine IoT devices to collect environmental data, and use smart meters to predict electricity consumption. Finally, we will construct a virtual world in the field of testing, and demonstrate the power-saving rules in a digitally-divided way, so that residents can predict whether electricity is safe in the future and the power-saving behavior that can be performed in the case of personal privacy protection.
