Detection&Location Of Unreported Leakage In Urban Water Supply Network

Leakage detection and location of urban water supply network is of great practical significance for saving water resources and ensuring the normal use of water by residents,enterprises and public institutions.However,some unreported leaks in the pipe network are not easy to be recognized and found.To solve this problem,the following research work is carried out in this paper in combination with the existing leakage detection research:1.Unreported leaks of pipeline network identification based on dynamic process.According to the characteristics of the evolution trend of unreported leaks,a targeted unreported leakage experiment is designed to reflect the process of developing from small leakage to large leakage.First of all,water supply network is divided into several partitions on the basis of node pressure sensitivity and K-means++ clustering algorithm.Then,hydraulic simulation software EPANET is used to conduct Monte Carlo stochastic simulation experiment of unreported leaks in example water networks.Last,a large number of simulation data are used to train the random forest classification model,and the nonlinear relationship between the unreported leak node and the corresponding leakage partition is established,so as to identify the area of the unreported leak.2.Community unreported leak detection based on accurate prediction of water demand.The temperature,weather conditions,solar terms,working days and holidays are set as specific integer code,which is used as input vector of QPSO-LSSVM community water demand prediction model to enhance the model's adaptability to changes in external conditions.On the basis of accurate prediction model is established,the difference between the actual water consumption and the predicted water demand as the input of CUSUM algorithm to detect the small leakage in the community network.At the same time the WEC method,another Statistical Process Control(SPC)method,is taking for comparison.The results show that the water demand difference analysis of CUSUM algorithm can effectively detect the community unreported leak,also large leak,and less number of false positives than WEC method.3.Leakage location based on improved PDLD method.Aiming at the problem that thetraditional PDLD method is difficult to determine the maximum number of leaks in the pipe network and the corresponding node emitter coefficient,a method of using flow analysis in the leakage period to determine the above two parameters is proposed.The leakage search range is reasonably narrowed by analyzing the degree of abnormality of the pressure measuring points in the pipe network.Finally,Symbiotic Organisms Search(SOS)algorithm is used to solve the PDLD model equation.Compared with the solution results of traditional genetic algorithm,the location accuracy of leakage points is improved,which has certain guiding effect on the actual network leakage location.