Study On Pipe Burst Positioning Model In Urban Water Supply Network Based On The Extreme Learning Machine

The water pipe bursting accident is a common phenomenon that often occurs in the water supply industry today.It is characterized by sudden bursts,serious hazards,and wide-ranging effects,which are likely to have a bad impact on urban production,residents’ living and environmental sanitation.With the development of social economy and the advancement of science and technology,the information technology of urban water supply pipe network has made a major breakthrough.It has great significance to conduct real-time pipe network leak detection and explosion for saving water resources and ensuring urban pipe network security through scientific means.Aiming at the blasting positioning,this paper puts forward a method which includes the hydraulic simulation of the squib accident,the optimal arrangement of the blast pipe pressure monitoring point and the blast pipe positioning based on the extreme learning machine.The main work is as follows:Firstly,a detailed analysis of a large number of squib historical data,the cause of the squib,the blasting accidents are classified according to the form of performance and the mechanism of the accident is explored,combined with the blasting records of various places over the years.Moreover,the occurrence and rules of blasting accidents are summarized.Secondly,according to the pipe network modeling theory method,the research and development of the hydraulic model of the pipe network is established and checked;for the case that the pipe network management department lacks the basic data for the real-time monitoring of the pipe burst,the pipe network model is used to simulate the pipe burst accident.Investigate the law of "burst-monitoring point pressure change",and propose the hydraulic simulation of the blasting condition by the application node flow multiple method,and verify the reliability of the method,which provides basic model support for optimizing the layout of the pressure monitoring points and the blasting positioning model.Thirdly,the existing pressure monitoring point layout method is analyzed.For the situation that the layout is unreasonable and the explosion pipe cannot be effectively positioned,a pressure monitoring point optimization arrangement method for the pipe burst accident is integrated.The pipe network is first partitioned into several areas according to the cluster analysis method.In the small area,the optimal arrangement of the pressure monitoring points is realized by calculating the influence coefficient matrix between the nodes of the blasting conditions and supplementing the Pearson correlation analysis.Finally,on the basis of the above research,the extreme learning machine algorithm is combined with the pipe network hydraulic simulation model,and the pipe network blasting positioning model is proposed.The model uses the learning and training mode of the extreme learning machine to simulate various hydraulic bursting parameters(burst position and squib flow)in the pipe network hydraulic simulation model under different working conditions.The bursting characteristic value corresponds to the data set of the position of the squib node,and the extreme learning machine model is constructed and trained,and finally the location of the squib point is located by the blasting characteristic value.When the blasting warning model based on the extreme learning machine is applied to the entire research pipe network area,the average prediction accuracy of the squib point is low.Therefore,it is further proposed to partition the entire pipe network and predict the squib in the small area,and set the number of different pressure points to compare the accuracy of the blasting positioning model.The results show that the accuracy of the pipe burst positioning model in the small-area pipe network is greatly improved,and the accuracy of the prediction increases with the increase of the number of pressure monitoring points in the area.However,when the number of pressure measuring points reaches a certain amount,the additional pressure monitoring points have limited help for improving the prediction accuracy.The relationship between the distance between the pulsation distance and the diameter of the squib and the prediction accuracy is found.When the position of the squib is close to the pressure point,the average prediction accuracy is higher,indicating that the sensitivity of the pressure point arranged based on the squib monitoring is good;the relationship between the diameter of the pipe burst and the prediction accuracy is further explored.With the increase of the pipe diameter of the pipe burst,the prediction accuracy of the explosion pipe positioning warning model is also increased,and the pressure is verified.The accuracy and rationality of the limit learning machine blasting positioning model for defining the blast characteristic value.In general,the water supply pipe network burst positioning model based on the extreme learning machine has strong operability in the daily management of the pipe network bursting,which can help the management personnel to narrow the detection range of the pipe burst accident point and reduce the detection difficulty.