應用MDMS資料之用電行為模式分析

An analysis of electrical load patterns and applications of MDMS data

隨著科技進步，越便利的生活則越增加對電的依賴，在尖峰時間用電負載量也隨之增高。唯有對用電行為有所了解，才能知道是否存在改善空間，以及如何規劃執行才能達到最佳的成效。本研究根據台電智慧電表在電表資料管理系統（MDMS）中累積的數據，提出每日用電模式的特徵擷取方式，並透過k-means分群方法進一步挖掘出隱藏的用戶慣常用電行為及行為改變模式等。透過此分析過程，電力供應者及電網管理者能夠更細緻地掌握用電變化至用戶層級，可藉此提出可行的策略改變用戶的用電行為以緩解可能的供電緊張情況，並增加電網營運上的利基。此外，也可藉由分析結果提供用戶個別化、差異化的資訊以增加用戶對自身用電行為的了解，並引起其進行節能節費的興趣，以作為節能措施實施推廣之助力。

Thanks to advances in technologies, life is getting more and more convenient. However, it also comes with problems of increasing dependence on energy, as well as high electric demands in peak hours. Therefore, analysis of user behavior is necessary for utility companies to find potential improvements and make better strategies. This study is based on the data collected from Taiwan Power Company’s Meter Data Management System (MDMS). A data processing method to extract features of daily electricity consumption pattern is proposed and k-means clustering method is used to observe hidden user behavior patterns and the factors of behavior changes. Through this analysis, utility companies and grid administrators can manage the amount of electric demand to the level of household and thus can make specific strategies to alter user behaviors for the purpose of better grid safety or operational profits. In addition, the results of this analysis can also provide individualized and distinctive information to users for their interests of understanding themselves’ consuming behaviors and motivations of energy conserving.