| Water is the source of life,with the increasing water consumption in industry,agriculture and domestic life,ensuring the sustainable use of water resources is a common problem faced by the water conservancy and water industry in all countries in the world in the development.The water supply pipeline system is the largest subsystem in the entire water transmission and distribution system,and the leakage of the water supply pipeline is the main reason for the loss of water resources in the entire water transmission and distribution system.The water loss caused by pipeline leakage in our country even exceeds the entire transmission and distribution system,that is 40% of the total water volume of the water system.Therefore,solving the leakage problem of the water supply pipeline system plays an important role in reducing the loss rate of water resources,improving the sustainable utilization of water resources,and building a green water-saving system.The main leak causes of water supply pipeline are aging,chemical corrosion,external damage and man-made damage,etc.Aiming at the problem of pipeline leakage control,acoustic signal leakage identification and location is one of the most widely used technologies.Due to the complex and dense distribution of urban underground water supply pipeline network,it is not a long straight pipeline.There are many non-straight pipe fittings such as branches,bends and valves in the pipeline network.When these non-straight pipe fittings interact with the fluid in the pipe,hydrodynamic noise is unavoidable.It is called non-straight pipe noise.This noise and the leakage sound signal belong to fluid noise,which can easily lead to the reduction of the accuracy of the sound signal leakage identification and positioning technology,and even the leakage identification and positioning work cannot be completed.Aiming at the interference problem of non-straight pipe fitting noise in the pipeline network system on leakage monitoring identification and leakage location,this article has the following main research contents and innovations.(1)In the research on the generation mechanism of leakage sound source and non-straight pipe fitting noise source,based on the theoretical basis of fluid mechanics and fluid dynamics,the turbulence characteristics,eddy current characteristics,and physical characteristics of cavitation phenomena in leaking pipes are analyzed,as well as the physical characteristics of non-straight pipe fittings.The physical characteristics of turbulence and vortex phenomena;based on the theory of fluid acoustics,the physical characteristics of turbulence,vortex-induced noise,and cavitation noise components in leaking sound sources in leak pipelines,as well as turbulence and vortices in non-straight pipe noise sources,are analyzed.The physical characteristics of the excited noise component.Computational fluid dynamics(CFD)numerical simulation technology is used to study the flow trend of fluid in leaking pipes and non-straight pipe fittings,and the acoustic boundary element method(BEM)numerical simulation technology is used to compare and analyze the leakage sound and the sound pressure level spectrum characteristics of the noise of non-straight pipe fittings.Finally,combined with the actual pipeline test experiment,the comparison and analysis conclusions of the sound pressure level spectrum characteristics of the two sound sources are verified.(2)In the research on the propagation mechanism of the leakage acoustic signal and the non-straight pipe noise signal,the leakage acoustic signal and the non-straight pipe noise signal are coupled to form the pipe mixed acoustic signal.The pipe mixed acoustic signal under the conditions of non-fluid-structure coupling and fluid-structure coupling are discussed separately.The propagation law of the pipeline mixed acoustic signal is derived,and the sound pressure expression of the pipeline mixed acoustic signal in the pipeline medium and the fluid medium is derived,which provides a physical mechanism basis for the propagation channel estimation of the pipeline mixed acoustic signal.To realize the propagation channel estimation of the mixed acoustic signal of the pipeline,a robust adaptive k-order cumulant blind equalization estimation(AKC blind equalization)algorithm is proposed to estimate the channel.The propagation channel of the leaking mixed sound signal is easier to converge than other blind equalization channel estimation algorithms,has stronger anti-interference ability and higher robustness.(3)In the pipeline acoustic signal leakage recognition algorithm research,based on the LST-KSVC(Least squares twin multi-class classification support vector machine),combined with the Gaussian mixture model(GMM)theory,an improved dual hyperplane multi-classification support is proposed Vector machine algorithm:GLT-KSVC(GMM least squares twin multi-class classification support vector machine)algorithm.The GLT-KSVC algorithm uses the GMM framework theory to assign classification weights to the leaked sample points,assigns a larger weight value to the center sample point,and assigns a smaller classification weight to outliers caused by the noise of non-straight pipe fittings.The GLT-KSVC classification algorithm reduces the sensitivity to outliers.Compared with the SVM,Twin SVC and LST-KSVC classification algorithms,the GLT-KSVC algorithm is not sensitive to outliers,so it can overcome the problem of outlier interference caused by non-straight pipe noise.The algorithm has higher generalization characteristics and classification the recognition accuracy is higher.(4)In the research of pipeline leakage sound signal localization method,in order to separate and extract the leakage sound source signal from the pipeline mixed sound signal,and solve the problem of the interference of the non-straight pipe fitting noise on the leakage sound localization,based on the E-Fast ICA blind separation algorithm,this paper proposes C-EFast ICA(Complex-valued E-Fast ICA)algorithm,C-EFast ICA can convert the convolutional separation model of the time-domain pipeline mixed acoustic signal into a linear product separation model in the frequency domain,and the channel estimation obtained by AKC blind channel estimation The function is used as the initial filling value of the separation matrix iteration of the blind separation process of the leaky mixed acoustic signal.Compared with the traditional method of randomly filling the initial iterative value of the separation matrix,the introduction of the channel estimation function makes the blind separation process of the leaky mixed acoustic signal stable and reliable.Compared with Fast ICA’s complex number domain expansion algorithm C-Fast ICA(Comlex-valued Fast ICA),the C-EFast ICA algorithm separates the mixed acoustic signal of the pipeline faster and has more separation,which greatly reduces the relative positioning error of the non-straight pipeline leakage point. |
PDF Full Text Request