Pipeline Leak Detection And Localization Based On Particle Filter

Pipeline is the main means of transporting oil resources.In recent years,the number of pipeline leakage accidents due to the aging of pipes and the drilling hole of oil stolen has been increasing,causing a large amuont of economic losses and casualties.Therefore,quick and accurate localization of pipeline leakage point is an urgent problem to be solved to ensure safe operation of pipelines.Firstly,the mathematical model of the pipeline is established,and the numerical solution of the model is obtained by using the characteristic and finite difference methods.In order to improve the accuracy of mathematical model,the influence of elevation and friction coefficient on pipeline fluid was considered.At the same time,as particle filter is used to estimate the state of the pipeline,the mathematical model is transfromed into the state space equations.TLNET was used to build a pipeline simulation platform to obtain the simulation experimental data of pipeline leakage and Matlab was used to perform numerical calculation based on the proposed pipeline mathematical model.The effectiveness of the mathematical model is verified by comparing the matching degree between the simulation data and the calculated data.Secondly,based on the established state space equations,the particle filter algorithm is designed to estimate the state of the pipeline.Besides,an effective method to estimate the leakage point is proposed based on secondary segmentation and particle filter algorithm.In order to improve the accuracy of leak localization,the idea of bootstrap fission particle filter is used to estimate the leakage quantity.The simulation results show that the proposed method,as a general or universal state estimation method,can detect and locate small leakage with relative error less than 1%.Finally,considering the low practicability of the state estimation method due to its high requirement on hardware,an improved pipeline leak localization method based on compressed sensing and particle filter is proposed.The principle of mass balance is used to detect pipeline leakage.The feasibility of compressed sensing in compression reconstruction of leakage signal and the influence of parameters of reconstruction algorithm on signal reconstruction are analyzed through a leakage signal measured in practice,based on which the performance of proposed leak localization method via particle filter is optimized.At the same time,the effectiveness and optimality of the proposed method are verified by practical experiments.The experimental results show that the measurement signal(100Hz)required by particle filter can be compressed by 65%~75%,and the relative positioning error of 0.95%~1.07% can be maintained while effectively locating the leakage at a number of different leakage points the leakage rates of which are within 2%~10%.