Study Of Water Distribution Network Burst Identification And Location

With the age of water pipe network increasing, more water resourse was wasted heavily beceause of frequent burst accidents and high water leakage rate.Water distribution network burst will also cause water head drops, secondary contamination of water quality, which will have effects on social, industrial and economic problems. So the identification and location of water distribution network burst has became a problem raised for urgent consideration and solution.This paper established Water Distribution Network Burst Probability Integration Model for single monitoring site, Water Distribution Network Burst Identification Model and Trend-surface Location Model of Water Distribution Network Burst. All of the models based on water distribution system SCADA database work together to relaize water distribution network burst identification and location.Firstly, LSSVR rained by pressure history was selected to estalished pressure prediction modet which were able to infer the probability of water distribution network burst for single monitoring site; K Nearest Neighbor Density was used to eatalish Pressure Tendency Identification Model to calculate the abnormal extent of pressure monitoring site behaviors;Adapituve Network Based Fuzzy Inferece (ANFIS) was traind by information got from pressure monitors to inference the probability of water distrubution network burst for single monitoring site.Secondly, an improved Kaverage Spatial Clustering was established with Genetic Algorithm and Information entropy theory to identify the working condition of water distribution network.The improved K average Spatial Clustering could overcome the current them, for instance, the outcomes from traditional K average Spatial Clustering were affected by initial cluster centers and weights.This paper establish Water Distribution Network Burst Identifation Model based on Weighted DS Theroy, the burst probability got from every pressure monitor site would be independent evidence for weight DS theroy Water Distribution Network Burst Identifation Model,working together with the improved K average Spatial Clustering.It was proved that the Distribution Network Burst Identifation Model based on Weighted DS Theroy could improved the certainty of identification.Thridly, based on predicted pressure values, mesured pressure values and water distribution Network monitoring coordinates,Trend-surface was established to position the water distribution network burst accident.The model was tested by water distribution network burst accident record. Extent of water distribution network burst was proved to be the foremost factor by analazing the Influences on the pinpoint accuracy.