Study Of Leak Detection And Diagnosis Method Of Chemical Process Pipeline

In the field of chemical engineering,the pipeline transporting fluid media plays an important role.As pipeline is affected by pipeline environmental conditions or the third parties,pipeline leakage often occurs.To reduce the influence of pipeline leakage on production safety and economy and ensure the stable operation of pipeline transportation,the early detection,location and maintenance of pipeline leakage are of great significance in theoretical research and engineering practice.With the development of sensing technology for pipeline inspection,the improvement of signal processing technology and the increase of pipeline project construction projects,the amount of data for pipeline inspection is also increasing.Therefore,how to process a large amount of measurement data is one of the hot research topics for pipeline leak detection and location.Based on reading a large amount of literature and summarizing the existing technical methods,this article has studied signal denoising,leak detection and leak location,etc.The specific work is as follows:(1)To reduce the influence of signal noise on leak detection and location,a method based on Improved Local Mean Decomposition(ISLMD)is proposed.Firstly,the cubic spline interpolation function is used to replace the median interpolation function in the Improved Local Mean Decomposition(ILMD)algorithm,and the local mean function and envelope estimation function are solved by the cubic spline interpolation function.Compared with the median interpolation function,the cubic spline interpolation function has better accuracy and efficiency.In order to solve the boundary distortion problem in ILMD,the method of boundary expansion is adopted.To solve the problem of boundary distortion in the calculation of the upper and lower envelopes,adaptive waveform matching technology and adaptively expanding the original signal are used.In order to increase the rationality of the selection of the Product Function(PF)after decomposition,a product correlation factor between the product function and the reference signal is calculated.The product correlation factor is used to determine whether the product function contains leaked information.The results show that the proposed method can effectively filter out the noise contained in the signal,restore the pressure signal at the upstream and downstream of the pipeline,and has a higher signal-to-noise ratio and a smaller root mean square error.(2)To solve the problems of lack of robustness and generalization of conventional leak detection methods,a pipeline leak detection method based on Convolutional Neural Network(CNN)is proposed,which can adaptively extract the leakage waveform features to detect different leakage sizes after converting the noise-reduced pipe measurement signal into an image signal.In addition,the CNN trained on a single point leak signal can also be extended to the detection of other leak points with high accuracy.The experimental results show that the CNN model used has better robustness and generalization ability.(3)To solve the problem of delay estimation for pipeline leak location,a method for pipeline leak location based on the amount of leakage to select weighting function of Generalized Cross-Correlation(GCC)is proposed.This method selects a weighting function based on the amount of leakage detected by the CNN.A SCOT weighting function is used in the case of small aperture leakage,while a PATH weighting function is used in the case of large aperture leakage.After selecting the weighting function,the generalized cross-correlation analysis of the pressure signals at the upstream and downstream is used to determine the time delay caused the leakage negative pressure and then locate the leakage.Compared with the use of a single weighting function,the selection of the weighting function according to the amount of leakage has better positioning accuracy.