Data-Based Detection And Location Of Pipe Burst Events In Urban Water Distribution System

Pipe burst detection and location is a hotspot in the safety production of urban water supply network.Many academics have done a lot of research on this problem.On the basis of analyzing the existing research,this paper carried out the following research:(1)Pipe burst detection based on the pressure difference SPC method: the pressure SPC method,the pressure difference SPC and the normal probability cumulative curve of pressure difference method were adopt to detect the pipe burst events with SX SCADA data.The experimental results showed that the pipe burst detection based on the pressure difference SPC has a higher detection ability,its false negative rate dropped to 0 and the false positive rate dropped to 56.26%.The reason was found that the SCADA pressure data is non-normal distribution,and the pressure difference data is normal distribution,and the SPC method is more suitable for normal distribution.(2)Analysis of the influencing factors of pipe burst detection: considering the pipe burst detection affected by a variety of factors,discuss the main factors(sampling frequency,burst flow and burst time,etc.)which may influence the pipe burst detection by using EPANETH2 to simulate pipe burst events.After the specially designed simulation experiments,it was concluded that the appropriate sampling interval is 1-2 minutes and the more smooth pressure of the node is,easier to detect.(3)Pipe burst location based on the centre-of-gravity method and the K-Means method: firstly,the pipe burst point was located roughly by clustering monitors,because it takes into account that the impact of the pipe burst on the surrounding nodes is not completely circular transmission.Then,the pipe burst point was located precisely by using the centre-of-gravity method on relevant monitors.The example verification shows that it is feasible to use this method to locate the pipe burst point and the locating error is between 50.2 m and 771.7 m.In conclusion,the pressure difference SPC method for pipe burst detection and the centre-of-gravity and K-Means method for pipe burst location are feasible in practical pipe network,and they could improve the pipe burst detection and positioning capabilities.