Study On Location Of Pipe Burst In Water Supply Networks Based On Model Driven And Mixed Statistics

With the continuous decline of data monitoring cost and the widespread application of hydraulic model of pipe network,the technology of pipe burst location based on data and model driven has been widely concerned.However,the positioning technology based on data and model driven is not only affected by the accuracy of monitoring data and model,but also related to the layout and number of monitoring points.Under the above influence,how to effectively improve the positioning accuracy of pipe burst based on data and model driven is a difficult problem.In order to solve this problem,this paper carried out the research on the location of water supply network burst based on model driven and mixed statistics.The specific research content and research results obtained are as follows:1)Research on the optimal layout of pressure monitoring points based on fuzzy clustering.The detection and location of pipe burst in water supply network need the data collected by monitoring points,so it is necessary to determine the number and location of monitoring points.Fuzzy clustering analysis is a kind of mathematical method that uses fuzzy mathematical language to classify things.It constructs fuzzy matrix according to the attributes of research objects,and uses fuzzy mathematical method to determine the fuzzy relationship between objects,so as to realize clustering.Based on the sensitivity analysis,the fuzzy clustering analysis method is used to optimize the layout of monitoring points in pipe network.Firstly,the pressure sensitivity matrix is calculated according to the Jacobian matrix derived by matrix analysis method.Then,the nodes are grouped by fuzzy clustering method,and the cluster center of each group is selected as the monitoring point.The results of case analysis show that the pressure monitoring points can be evenly distributed in the pipe network according to the topological structure after the optimized layout,which is conducive to improving the positioning accuracy of pipe burst.2)Research on the influence of background noise on the location of tube burst based on model driven.Based on the steady-state hydraulic model of pipe network,the model driven positioning of pipe burst is realized by analyzing the changes of pressure,flow and other parameters caused by pipe burst according to the real-time monitoring data collected by SCADA system.Considering that the monitoring data error is the main factor affecting the performance of the model driven method,the influence of background noise on the model driven method is studied.Taking two different scale pipe networks as an example,the positioning performance of the model driven method under the influence of different levels of monitoring error is compared.The results show that: when there is no background noise,the false alarm rate of the model driven method is 27% and 24%,while when considering the background noise,the false alarm rate of the model driven method is as high as 72% and 76%,which indicates that the background noise can significantly reduce the robustness of the model driven method for locating pipe burst.3)Study on Location of Pipe Burst in Water Supply Networks based on Model Driven and Mixed Statistics.In order to improve the positioning performance of model driven method under background noise,a study on the positioning of pipe burst in water supply network based on model driven and mixed statistics was carried out.Hybrid statistical process control method is a statistical analysis method combining Kalman filter and cumulative sum algorithm.It can identify and amplify abnormal signal from background noise,and then improve the performance of pipe burst detection and location.In this paper,EPANET and Water GEMS software are used to simulate the hydraulic state of small pipe network,science and Technology Park pipe network and c-town pipe network respectively,to clarify the implementation process of hybrid statistical process control method,to compare and analyze the robustness of Kalman filter and hybrid statistical algorithm,and to demonstrate the feasibility of hybrid statistical algorithm.The results show that: under the influence of background noise,the model driven method can only narrow the range of pipe burst points,and can not accurately locate the pipe burst nodes;Kalman filter has a certain filtering effect on the background noise,but still can not accurately locate the pipe burst position;the hybrid statistical algorithm is highly robust,and combines the burst warning and burst location technology to quickly locate the burst node,reducing the rate of false alarms during the non-burst period.