請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/57880
標題: | 多代理人感測網路於建築物內需求導向空調節能之研究 Demand-Driven Power Saving for HVAC System with Agent-Based Sensor Network in Buildings |
作者: | Yi-ting Tsao 曹怡亭 |
指導教授: | 許永真 |
關鍵字: | 節能,中央空調,多代理人, Power Saving,HVAC,Multi-agent, |
出版年 : | 2014 |
學位: | 博士 |
摘要: | 從統計資料中得知,依據地區的不同,有 20% 至 50% 的建築物內用電量是來自於中央空調系統。因此在環保與節能廣受重視的今日,降低中央空調系統的耗電量成為大家所關注的議題。本研究旨在依據需求來改善建築物內能源管理系統之效能,以台灣大學資訊工程系館的中央空調系統作為標的,測試各項改善方法的成效。
建築物內能源管理系統中,可依功能大略分為量測、控制與節能機制。我們架構基於多代理人的感測網路來提供統一的存取介面,以降低結合不同感測器資料的難度。另外,我們亦提出估計中央空調需求的方式,進一步將此一概念用於控制以及節能機制設計上。在控制層面,我們比較三種不同的控制方法在使用者舒適度、用電量以及系統穩定性上的效果。在機制設計層面,我們以需求做為電費分攤依據,並以需求進行電價的浮動調整,以期能克服由於不同環境條件所造成的不公平性。 透過問卷調查指出,基於多代理人的感測網路可在學習存取感測器上降低時間於原來的三分之一。而模擬實驗結果指出,需求導向的控制方法在不影響舒適程度的狀況下,可於日間節省 1.74% 至 6.51% 的用電量。此外,我們發現預算機制在以需求為電費分攤依據下,有助於降低尖峰時刻的用電量。 Heating, Ventilation, and Air Conditioning (HVAC) systems account for 20% to 50% of power consumption in buildings. Reducing the power consumption of HVAC systems has become a popular issue with the growing concerns for energy conservation. This research focuses on demand-driven performance improvement of Building Energy Management Systems (BEMS)for the HVAC system in the Department of Computer Science and Information Engineering (CSIE) building at National Taiwan University (NTU). Generally speaking, the functionality of BEMS consists of measurement, control, and power saving. This thesis proposes using cooling demand estimation as the basis for designing the energy saving mechanism. Three approaches to control are compared on human comfort, power consumption, and maintenance. The experimental results showed that the demand driven control approach saved 1.74% to 6.51% of power consumption in the daytime. To overcome the problem due to climate differences, we apportion the electricity payment on demand by adjusting its price accordingly, We observe that demand-based apportionment may help reduce the power consumption during peak hours. To mitigate the difficulty in fusing heterogeneous sensor data, an agent-based sensor network is constructed to provide an integrated interface for data access. A user study is conducted to show that the agent-based sensor network reduces data access time by about one third. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/57880 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-103-1.pdf 目前未授權公開取用 | 10.81 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。