Experimental Study On Locating Leakage Point Of Water Supply Pipeline Based On Acoustic Signal Analysis Of Leakage

Urban water supply is an important sector of national economy and national health,pipeline leakage is an important issue which threat to water supply,a variety of natural and human factors such as geological subsidence,soil corrosion,construction,operation and management problems will cause pipeline leakage,such as water supply pipelines buried underground,more leakage is not easy to be found,not only caused the waste of resources and energy,are more likely to cause surface collapse,water pressure,water quality decline threat to water supply such as serious accidents.In a variety of pipeline leak detection method,the acoustic method is relatively common,such as correlation analysis,listen to leakage tester,etc.,but these methods need skilled workers experience and strict condition of pipeline,the paper explore a new method of pipeline leakage localization,the research on the ground to collect leakage acoustic signal,noise reduction,to extract characteristics of acoustic signal processing,such as substitution based on BP neural network model to achieve the purpose of identification of leakage location.Firstly,the influence of pipeline leakage point on soil erosion was studied,and the influence of the size and orientation of leakage hole on the formation of erosion pit was explored.It was found that the size of erosion pit would gradually stabilize with the passage of time,and the size of erosion pit increased rapidly when the leakage hole was large and the leakage hole was oriented upward.By simulating actual water supply pipeline,on the ground to collect different pressure,leakage mouth size,leaks toward,pipe leakage acoustic signals under different conditions,such as,the change law of various characteristics under different conditions,found the tube pressure reduces,covering thickness increases,the mouth size is reduced,leakage acoustic signal gradually reduce the value of the power,the discrete degree increased gradually;Pipe and surface covering materials have great influence on the characteristics of acoustic signals.When the leakage port is facing down,the power value of the leakage acoustic signal is smaller than that of the leakage port facing up and right,and the randomness and complexity are also higher.Secondly,based on the time-frequency characteristics of all kinds of noise on the pipe outside noise classification analysis,will be used in several common speech signal noise reduction methods such as spectrum subtraction,wiener filtering method was applied to leakage acoustic signal,the comparative analysis of several noise reduction methods on different SNR(0,2,5,10 db)of the signal with noise,the noise reduction effect of various kinds of noise reduction methods,were treated with method of wiener filtering signal SNR by up,generally more than 4 db,in low SNR,especially reached 9 db,the effect of noise reduction method comprehensive sorting as follows:Improved spectral subtraction method of wiener filter > multiwindow spectral estimation >Boll improved spectral subtraction method > basic spectral subtraction method.When the signal-to-noise ratio is lower,the signal contains more noise components and the noise reduction effect is better.Finally,based on BP neural network to identifying and localizing the leakage signal model is set up,collecting all kinds of different conditions of leakage acoustic signals and ambient noise,the leakage acoustic signal noise reduction processing,extraction of leakage signal and many kinds of characteristic value of environmental noise as neural network input layer,leakage probability as output layer to establish BP neural network identification model,the model in the laboratory simulation of pipeline leakage judgement probability majority can achieve 80% above,near to the actual water supply pipeline suspected funnelled to leakage points identification and positioning error is within the range of 1 m.