随机流量供水管网漏损诊断阈值及误差估计

实际供水管网的漏损和流量随机性均会导致管网水力分布的波动,水压监测中通过压差阈值界定两种状态,但传统阈值只能控制误报率,不能控制诊断误差。为降低漏损诊断误差,在监测目标明确的情况下,给出对应最小误差的诊断阈值,并将其应用于Lansey供水管网中。结果表明,在随机流量供水管网中,该阈值稳定性好,对应的漏损诊断精度高;由于得到的漏报率和误报率接近,实际管网漏损诊断中,可以采用该误报率的2倍作为漏损诊断误差。

Leakage Diagnosis Threshold and Error Estimation of Water Supply Network with Random Demand

In practical water supply network (WSN), water demand is time varying and random. Both of leakage and random flow lead to the fluctuation of hydraulic distribution in WSN. The diagnose threshold of pressure drop is set to define two different states. The traditional diagnostic threshold can only control the false alert ratio and not control the diagnosis error. In order to reduce the leakage diagnosis error, the diagnosis threshold based minimum diagnosis error was presented under a predetermined goal of leakage detection. The method was applied to Lansey WSN. The results show that the threshold has strong robustness, and is relate to higher accuracy of leakage diagnosis. Because the false alert ratio is close to the missing report ratio, two times of the false alert ratio can be used as the leakage diagnosis error in practical WSN.