Research On Factors Of Pipe Bursts And Location Method For Urban Water Supply Networks Based On Artificial Neural Network

Urban water supply network is an important infrastructure in cities, which affects the social life in many ways. However, pipe bursts, which happen frequently, have harmed the safety of city water supply, which have drawn attention of scholars and technicians. From digging up the factors and seeking the law of pipe bursts, maintenance warnings can be launched in advance, which will lower down the frequency of the pipe burst accidents. Meanwhile, harms from accidents of pipe bursts can be eliminated effectively and handling capacity of water supply department can be improved from the development of quick location techniques for pipe bursts. These two works mentioned above have important research values on the work of saving water resource, ensuring water safety in society and improving economic benefits. The laws of pipe bursts of Fangcun district in the city of Guangzhou are researched, and quick location techniques of pipe bursts have been researched based on BP neural network.Firstly, main factors and the laws of pipe bursts are analyzed: the probability of pipe bursts of large radius pipes are lower than that of small radius pipes; the probability of pipe bursts of steel and cement pipes are lower than that of common iron pipes and plastic pipes; non-human pipe material factor is the main factor of pipe bursts; Slow lanes and sidewalks are places that pipe bursts happens easily. Otherwise, centralization of the law of accidents of pipe bursts are reflected from drafting the space and time law chart of pipe burst points: accidents mainly happen at places that have complex link structures and the months which have low temperature.Secondly, analytic hierarchy process is used to establish multi hierarchy weight analysis model to quantify the law of accidents of pipe bursts, the results are as follows:material of pipes0.5081, installation of pipes0.2161, vehicle dynamic load0.1355, radius of pipes0.0947, destroying events0.0457.Finally, procedure of quick location of pipe bursts is studied using monitoring data of pipe networks, BP neural network as the core. According to the study of law of pipe bursts and experimental contrasts, the neural network location model is established in the form of joint of pressure and time parameters, and optimal combination of functions are selected through inner function orthogonal experiment: training function trainlm, convergence function learngdm, transfer function logsig. Through simulation experiment checking, the effect of location can reach to:when the confidence region is0.75%of the objective region, the precision of prediction of the procedure is85%.The study results above can provide practical help for analyzing and controlling accidents of pipe bursts of city water supply network, and offering scientific accordance and technology support to the safety and management of city water supply systems.