請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/48881
標題: | 台灣住宅部門冷氣負載用電行為探勘分析研究 The Application of Data Mining to Air Conditioner Usage Behavior Research in Taiwan |
作者: | Hsueh-Chia Lee 李學甲 |
指導教授: | 曹承礎 |
關鍵字: | 資料探勘,冷氣密度,冷氣用電行為,決策樹,電力負載, data mining,power density of air-conditioner,air-conditioner usage behavior,electricity load pattern,decision tree, |
出版年 : | 2015 |
學位: | 碩士 |
摘要: | 近年來,全球能源供需問題日益嚴重,能源價格上升、氣候變遷、電子產品及交通工具增加,均使能源消耗更為嚴重,台灣也不例外。電力產業為一特別的產業,具有即產即銷的特性,必須達到電力即時供需平衡才不會造成生產過剩或供電不足。為達電力供需平衡可由電力供給面及電力需求面著手。從電力供給面來看在台灣目前受到許多因素限制,地理腹地小、環保意識抬頭限制建立核電廠、及九成以上能源供給仰賴進口受國際情勢影響等因素,都是電力發展從供給面著手經常會受到的阻礙。需求面管理則是另一個可行的方法,透過調節需求面對電力的需求,延緩對於電力供給面的依賴,提供負載管理策略,縮短尖離峰負載差距,達到節約能源效果等,也都能有效降低對整體電力的需求。
有鑑於發展電力需求面管理之重要,本論文旨在於探求台灣住宅部門電力需求面管理的節電潛力,本論文針對台灣住宅部門冷氣負載用電行為進行探勘分析研究,透過資料分析及探勘技術,挖掘住宅冷氣用電可抑低負載的潛力和空間。本研究採用台電公司97年度家電調查之問卷原始資料及台電公司用戶電表資料進行分析及探勘。首先,以資料前處理進行資料清理、整合、創造衍生性欄位,創造出冷氣用電密度、冷氣用電時段類型、每日冷氣用電度數等具代表性的欄位。再者,從冷氣密度中,我們發現冷氣密度出現明顯集群,本論文將用戶區分成五群不同等級之用電密度,並透過卡方及統計檢定驗證各群之間能有效區隔不同用戶。為了進一步了解各群內部特徵,我們以決策樹開展各群用電行為分為19大類,並了解每一類用戶之組成及其用電行為的內涵。 最後,我們依照19大類提出四個構面、六點需求面管理策略建議,包含階梯電價、冷氣中低用電量者節能回饋、夜間冷氣電價、直接/間接負載控制、優惠誘因措施、健康節能宣導等。希冀透過本論文研究,挖掘冷氣用電可以抑低負載的潛力和空間,協助電力公司提高住宅部門負載管理的效能,增進台灣整體電力需求面管理,達到節能減碳,穩定國家用電之終極目標。 In recent years, Energy crisis has been a global issue and getting worse on every second. Because of growing energy prices, increasing usage of electronic products, global warming, and climate change makes the supply could hardly meet the demand. Taiwan is no exception. To achieve power balance between supply and demand, we can process on the power supply side and the power demand side. From the perspective of power supply in Taiwan, Taiwan encounter a number of factors limitation, such as small hinterland, environmental consciousness for build a nuclear power plant, and over ninety percent energy supply imported aboard which may influence by international factors. Demand-side management is another possible solution, through adjusting the demand for electricity demand, shorten the gap between the peak and off-peak load through load management, and by saving energy can also reduce the overall demand for electricity. This thesis aims to explore the energy-saving potential of Taiwan's electricity demand side management in the residential sector, and analysis Taiwan household air conditioning electricity usage and load pattern through statistical analysis and data mining technology. Our data base on the 2008 Taipower appliances survey questionnaire. First, by data preprocessing we create derivative field power density of air conditioners, air-conditioning power usage period type, consumption of daily air-conditioners and other electrical field representatives. Through the density of air-conditionings, we found significant clusters and divided user into five density groups, verified by Chi-square test and statistical verification. To further understand the characteristics of each cluster, we use decision tree on each group and divided into 19 categories of electricity behavior for understanding the behavior of air-condition usage and behavior. Finally, we present four dimensions, six demand side management strategy recommendations base on 19 categories, including level pricing, low power consumption rebate, night time air-conditioning pricing, direct/indirect load control, incentives, energy-saving health advocacy and so on. Hoping through this thesis, we can figure out the potential to save energy through air-conditioning usage and assist power companies to improve the efficiency of the residential sector electricity load management, enhance Taiwan's overall electricity demand-side management, to achieve energy saving and carbon reduction. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/48881 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊管理學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-104-1.pdf 目前未授權公開取用 | 4.5 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。