ABSTRACT

The spatiotemporal correlations typically exist between the monitoring data provided by the devices distributed at the water distribution system (WDS). The leakage detection method featured by the utilization of the spatiotemporal correlation of the monitoring data had been proposed in our previous study and parameter settings have a significant impact on the detection performance of the method. This paper focuses on the optimization of two key parameters of this method, Scanning Time Window Length (STWL) and Rate of False Alarm of the k^th scan (RF_k). STWL determines the length of the time-series data participating in the correlation analysis, and RF_k determines the weight of each scan in a complete scan process. The proposed method is applied in a realistic and full-scale water distribution network with the consideration of the demand uncertainty, leakage flow rates, and sensor noise. The results show that the optimized parameter settings could improve the leakage Detection Probability (DP).

摘要

时空相关性通常存在于在水分配系统（WDS）处分配的设备所提供的监控数据之间。在我们先前的研究中已经提出了以监测数据的时空相关性为特征的泄漏检测方法，参数设置对该方法的检测性能有重要影响。本文重点研究该方法的两个关键参数的优化，即扫描时间窗口长度（STWL）和第k^次扫描的虚警率（RF _k）。STWL确定参与相关性分析的时间序列数据的长度，RF _k确定整个扫描过程中每次扫描的权重。在考虑需求不确定性，泄漏流量和传感器噪声的情况下，将所提出的方法应用于现实的，大规模的供水网络。结果表明，优化的参数设置可以提高泄漏检测概率（DP）。