都市水資源管理的重要性日益提升，為了協助維運管理及分析決策，水資源管理單位亟需推動適當工具的應用。以自來水管線為例，期望運用人工智慧、水理分析及其他技術，幫助決策者了解供水分區內水壓及流量的整體分佈和可能的漏水情況，進一步分析各管線的風險特性，並針對高風險管線提出改善及永續發展策略。本研究導入水理分析於風險評估中，以克服過去在風險分析中無法精確計算受影響用戶的缺點，提供更優化的風險評估建議。 本研究分三個階段進行，第一階段：以臺北自來水事業處現有自來水管網架構為基礎，根據水理分析應用需求，自動化轉出水理分析所需的管網結構及INP檔案。其轉製成果除了能支援水理分析軟體(EPANET)的讀取及分析，還與商業軟體(WaterGEMs)比較，其相似程度超過99%，證明自動化匯出具參考價值。第二階段：採用遺傳演算法進行模型校正作業。經過校正程序調整參數後，分析結果與現地量測結果趨近，且校正成果與WaterGEMs比較，更為良好。此外，與監控點進行24小時比對，誤差僅為8.97%。第三階段：以羅吉斯迴歸與先進的水理分析技術，結合液化潛能的深度評估，開發出一個創新的整合式風險評估模式 (SynerRisk)，這不僅能動態量化自來水管線可能影響的用戶規模，還能將複雜的分析結果精準反饋到每條管線，提供專為臺北自來水事業處量身定制的管理方案。這一模式不僅能大幅提升維護管理效率，還可為管線汰換決策提供可靠的數據支持，從而實現精細化管理。 ;Water resource management is a core method for ensuring the world’s sustainable development. Water resource management units urgently need to promote the use of appropriate tools to assist in maintenance, operational management, analysis, and decision-making. This study anticipates leveraging artificial intelligence, hydrological analysis, and other technologies to aid decision-makers in understanding the overall distribution of water pressure and flow within the water supply zone, as well as identifying potential water leakage conditions. Further analysis of the risk characteristics of each pipeline will allow us to propose strategies for improvement and sustainable development, mainly focusing on high-risk pipelines. This study integrates hydrological analysis into risk assessment to provide a clearer understanding of the potential impact on users, addressing past shortcomings by enhancing the accuracy of calculating affected users in risk analysis, thereby offering more optimized risk assessment recommendations. The findings of this research are categorized into three stages. 1.Initial Stage: This stage utilizes the existing water pipeline network structure of the Taipei Water Department and automates the extraction of the pipeline network structure to generate the required INP files for hydraulic analysis. In addition to facilitating the reading and analysis by hydraulic analysis software (EPANET), the conversion results are compared with commercial software (WaterGEMs) and show a similarity of over 99%WaterGEMs, indicating the reliability of the automated export process. 2.Second Stage: This stage involves employing genetic algorithms to conduct model calibration. After adjusting the model parameters through the calibration procedure, the analysis results closely align with on-site measurement results. The calibration results are superior to those obtained with WaterGEMs, the error of only 8.97% compared to 24-hour monitoring data. 3.Final Stage: This stage combines logistic regression, hydraulic analysis, and liquefaction potential to conduct risk assessments of water pipelines. The integrated risk assessment model (SynerRisk) accurately depicts the number of affected users and provides detailed analysis results for individual pipelines. This approach better meets the management requirements of the Taipei Water Department and serves as a foundation for pipeline maintenance management and replacement decisions.
