自來水管網漏水率改善，過度依賴固定資產管線的汰換率，則需投入大量預算，然在政治、氣候及經濟影響因素下，無法調整水費收入，進而增加預算進行管線汰換下，突顯主動式漏水檢測作業重要性，即以最經濟的方法進行漏水管線重點式的改善，而非區域全面汰換或屆齡汰換，目前文獻上漏水預估模型分類，小區計量模型、水力模型及漏水潛勢模型，因受限於條件限制下，實務上，不易整體性的運用於自來水管網檢測作業，目前漏水檢測作業多以供水系統漏水率嚴重之區域進行重點式巡檢，其成果與效率受限於檢測人員的經驗與經歷，無法達到客觀性判斷及預知漏水點之目標。
有鑑於此，本研究以市區巷弄之街廓為分割單元，蒐集自來水管漏水因子，經資料前處後，首先以羅吉斯迴歸(Logistic Regression)導入修正街廓管線漏水機率，並利用決策樹分析（Decision Tree, DT）進行漏水預測，接著利用將「路面漏水」、「檢測地下漏水」資料庫，依據壓力管線先漏後破理論（Leak Before Break, LBB），將二種不同狀態漏水點作空間相關性分析，最後結合前項決策樹漏水預測結果，顯示本研究由決策樹架構，配合導入羅吉斯迴歸修正及空間相關性所建決策樹模型具有合理性與可用性，可作為日後供水管網系統漏水診斷推估發生漏水之街廓及排定檢測漏水路段順序。

Over-reliance on fixed assets pipeline replacement for improving water distribution network leakage rate condition, water utilities have to invest a huge of budget. However, under water consumption policy, climate, and economic factors, water revenue is hard to increase. Thus, under the circumstance that the budget for pipeline replacement is limited, a proactive leak detection operation is the most economical method for water pipeline leakage improvement. Rather than completely replaced the whole aged pipe in an area, some previous researches propose leakage detection models to find leakage hotspots in the network. These including: break down large water network to small DMAs, residential econometric model, water hydraulic simulation model. Due to limited conditions under the substantive and difficult to apply water pipe network integrity testing operations, the current leakage inspection and detection only focus on serious leakage areas, its effectiveness and efficiency is limited and depend on the experience of testing personnel.
For this reason, this study alleys of urban street blocks as a split unit for collecting water pipe leakage factors. After data processing, first with logistic regression (Logistic Regression) introducing amendments to street blocks pipeline leak probability and then use Decision Tree Analysis (DTA) to predict leak. After that, the use of the "road leakage," "detect underground leaks" database, according to the pressure line leak-before-break theory (Leak Before Break, LBB), will be two different states leakage points for the spatial correlation analysis. Finally, combining into a leak-before-break DTA forecasting model for leak prediction. The results of this study include the tree structure, with the guide Logistic regression correction and spatial correlation of the decision tree model was built and is reasonable availability, for future water supply system leaks occur street blocks estimate leakage of diagnosis and detection of leaking sections in a scheduled order.

1.Thomas M. Walski(2009)，Detection of Patterns in Water Distribution Pipe Breakage Using Spatial Scan Statistics for Point Events in a Physical Network.
2.Wallace, L.P(1987)，Water and Revenue Losses： Unaccounted for Water. American Water Works Association Research Foundation, Denver, CO.
3.Zheng Yi Wu(2005)，WATER LOSS DETECTION VIA GENETIC ALGORITHM OPTIMIZATION-BASED MODEL CALIBRATION.
4.Milind Naphade(2011)，Smart Water Pilot Study Report.
5.Larry W. Mays(2000)，Water Distribution Systems Handbook. McGraw-Hill.
6.James A. Ligget and Li-Chung Chen(1994)，INVERSE TRANSIENT ANALYSIS IN PIPE NETWORKS.
7.Manual of Water Supply Practice(1999)，Water Audits and Leak Detection. American Water Works Association.
8.IWA(2002)，Water Losses Management and Techniques，INTERNATIONAL REPORT.
9.Pearson(2005)，How does Pressure Reduction Reduce Burst Frequency.
10.Cyprus(2005)，Experiences in DMA redesign at the Water Board of Lemesos, 2005.
11.Kenneth(2003)，THE IWA WATER LOSS TASK FORCE (Water 21 - Article no. 1) Practical Approach Initiatives to Water Loss Reduction.
12.莊立偉 (2011)「自來水測漏小區管網劃分程序及優選模式」，交通大學環境工程研究所碩士論文。
13.張明儒,杜炎勳,蔡柏軍(2011)，核能電廠圓周向穿壁裂縫管線之J-積分計算，原子能委員會核能研究所。
14.張明智(2005)，「以空間資料搜尋支援資料探勘之研究」，國立成功大學測量及空間資訊學系碩士論文。
15.何承嶧( 2009)，「自來水配水管網漏水現象之分析、預測與管理之研究」，國立臺灣大學工學院環境工程學研究所博士論文。
16.駱尚廉、林文淵、陳亦能、楊昆霖、廖浡廷(1997)，「減少漏損及無費用水之經濟研究」。
17.何承嶧、林明德、駱尚廉(2009)，「運用不同預測模式推估自來水漏水事件」，漏水防制與管理研討會論文集。
18.日本水道協會(1998)水道維持管理指針。
19.楊境維、黃欽稜(2009)，「臺北自來水事業處主動漏水控制與管理作為」，漏水防制與管理研討會論文集。
20.廖述賢、溫志皓(2000)，「資料採礦與商業智慧」，雙葉書廊。
21.程爱玲(2007)，城市供水管網漏损預測模型及檢漏方法的研究。
22.張宏偉、牛誌廣、陳 超(2001)，供水管道漏損預測模型研究。
23.田林，張宏偉(2004)，應用自控閥控制給水管網漏損的研究。
24.耿為民、劉遂慶、信崑崙(2004)，給水管網漏損檢測週期的優化求解。
25.傅玉芬(2004)，城市供水管網漏損控制。
26.趙洪賓、陳兵、伍悅濱(2001)，給水管網漏失預測模型的研究。
27.中華技術(2014)，臺北盆地三維防災地質資料庫建置。
28.臺北自來水事業處(2014)，管線委外檢測工作成果報告書。
29.IWA(2014)，國際供水服務統計報告。