Reseach On Leakage Dischage Quantization Of Pipe Networks With Multi-Leak Sourses Based On Algorithm Of Improved ICA-R

With the rapid development of economy and growth consumption of the living water and industrial water in our country,pipe network has been under huge pressure and overstretched,leakage becomes the general phenomenon of water supply network,and leakage rate of water supply network is on the rise in recent years,leakage caused the huge waste of water resources and energy.Firstly,work for leakage reduction of water supply network depends on being in know the leakage level of water supply network,however,water supply network is buried underground and there is much difficulty for measurement.It is the direction of the research to get the background leakage value through the pressure and flow data of the on-line monitoring of the district pipe network by making full use of the Internet of things technology to adopt the scientific and reasonable signal processing algorithm.Based on the theory of blind source separation,the article researches on leakage-quantitative for pipe networks with multi-leak sources.Firstly,with establishing a numerical simulation of the pipe network structures as example by EPANET and leakage test platform for experient,observation signal composed of inlet flow and inlet pressure was obtained by experimental project which collected inlet flow and inlet pressure as observation signals for different number of leak sources,and the actual water consumption and leakage as the source signal used to verify the separation effect.The signal provides data to support the feasibility of blind source separation theory and verifying the non-gaussian of source signals provides theoretical support for blind source separation.Secondly,analyzing blind source separation theory to determine the objective function and establish a quantitative model for leakage.Through analyzing the disadvantage of existing algorithm to screening the appropriate algorithm,an improved algorithm ICA-R is proposed which is based on the CICA;The conventional algorithm and improved algorithm are applied to the leakage quantitative model for blind source separation,it is used for signal separation according to data from the pipe network example and experiment platform and get separation trends which compared with the trends of source signal.Result of separation is evaluated by Pearson similarity coefficient to prove that the waveform separable information by ICA-R is accurate over 95%.Finally,choose good trend of separation effect to amplitude reduction and get real restore value.Through the average relative error of evaluation under 15% which makes error analysis with real source signal,it proved that improved ICA-R algorithm is more appropriate for leakage quantization of pipe networks with multi-leak sources and has better system noise immunity.