The Study Of Soft Computing For The Complex Water Supply Pipe Network Leakage Fault Diagnosis Based On Hydraulic Transients

The sustainable development of national economy and people’s livelihood and socialeconomy are tied up with the safety and reliability of the urban water supply. In view of theurban water supply network is a huge, complicated pipeline, and the line contact betweencomplex system influence each other, it has extremely important practical significance todevelop the water supply network of leakage fault diagnosis, to research and establish thepipe network leakage fault detection and positioning system with reliable and highaccuracy.The research is mainly about leakage detection and localization in municipalitysupply network based on the hydraulic transient.First of all，the actual state of water resource and leakage in municipal water supplynetwork, the reason of leakage, the evolvement of research on leakage including equipmentand the way of leakage detection are introduced. Contrapose the research state of waternetworks leak detection model in our country, a new dynamic signal prediction diagnosismethod based on soft computing with the function of a magnifying glass is put forward.Secondly, as the situation of very little historical materials and data of the accidentpipe network, a mathematical simulation method of the bursting accident of water pipe isintroduced,On this basis, a method based on the acuity and discrete combination optimizationarrangement model of network monitoring with the improved genetic algorithm ispresented. Compare with the results of the pressure monitoring points, the results of flowmonitoring points after optimization arrangement can effectively monitor the burstingaccident of the measuring point, at the same time, it can also lock the regon of burstingaccident as accurate as possible.Third，this model is established with least squares support vector machines (LS-SVM) by input data normalization and binary output data, and used the radial basis kernelfunction, at the same time, the particle swarm optimization (pso) algorithm is adopted tooptimize regularization parameters of least squares support vector machine (SVM) andradial basis function (RBF) parameters optimization, so as to ensure the effectiveimplementation of the mapping model diagnosis, to achieve the purpose of forecastingmodel.Finally, on the basis of the above work, this paper adopts from monitoring stations, the spectrum data, which comes from monitoring stations, after the Hilbert-Huangtransform analysis, is used as the learning sample of PSO-SVM model. By two differentbursting scale samples training, a computing intelligence bursting prediction and diagnosissystem with the function of magnifying glass is established, with which synchronousdiagnosis and prediction of the location and degree of leakage. The research proves that nomatter compared with static signal method nor simple generalized model, the complexwater supply pipe network leakage fault diagnosis soft computing based on the hydraulictransient has much higher diagnostic accuracy in locking the location and confirming thedegree of tube blasting in pipe network.The research for the municipal pipe network leak detection and control provides apractical way. Furthermore, the basis and methods of support is provided for the urbanwater supply systems safe running and management.