- 學位論文
台灣住宅部門冷氣用電行為分群探勘研究
The Clustering to Air Conditioner Usage of Household Behavior Research in Taiwan
摘要
由於電子智能、自動化的產品蓬勃發展，不管在工業或民生方面皆提升電力的依賴性與重要重要性。然而隨著環境、核能議題與人民的環保意識提升，加上電廠建置速度趕不上設備折舊速度、電廠建置成本等，電力供給面困難度卻日益提高。 為了改善台灣電力供需不平衡的問題，本論文與台電專案配合，利用台電102年度家用電器普及狀況調查表的用戶問卷資料、台電帳單系統、以及政府開放資料，針對住宅部門中的冷氣用戶進行分析，篩選出關鍵代表變數：居住坪數、冷氣用電量、7,8月與5,6月用電比、用戶年所得、房價排名。並以問卷的用戶作為訓練資料，先透過建立母體變數與樣本變數的關聯，再將五個關鍵變數透過K-means模型進行分群。因參考輪廓係素最大值，故將K設定為7，將用戶為7種類型。再一一針對各群描述用戶輪廓，並提出對應的冷氣節電建議，作為台電擬定節電策略的參考。
並列摘要
In order to improve the imbalance between power supply and demand in Taiwan, the paper cooperates with the Taiwan Power Company project to make use of the user questionnaire data, New Billing System, and Government Open Dada for air-conditioners in the residential sector analysis. Screening out the key proxy variables: residential Ping, air-conditioning electricity consumption, month electricity ratio, the user's annual income, house price ranking. Using the questionnaire users as the training data, the five key variables were grouped by K-means model by first establishing the relationship between the population variables and the sample variables. Due to the Silhouettes coefficient maximum, so K is set to 7, the users are clustered for the 7 types. One by one for each group to describe the user profile, and propose the corresponding air-conditioning energy-saving recommendations, as a reference for the Taiwan Power Company proposed power-saving strategy.