Research On Leakage Detection Technology Of Natural Gas Pipeline Combined With VMD And SVM

As the main means of transportation of natural gas and other energy materials,pipelines are indispensable in ensuring people’s normal production and life.However,due to the insufficiency of pipeline manufacturing technology,aging and corrosion caused by long service time of pipeline and the illegal behavior of stealing gas and oil,pipeline leakage will occur.If not found in time and take corresponding measures,it will cause a waste of resources and economic losses,and even threaten people’s lives.Leakage detection is the main technical means to ensure the normal operation of pipeline,and the acoustic signal generated by the natural gas pipeline leak is vulnerable to the interference from the external environment during the detection process,which leads to the occurrence of false alarm and missing alarm.Therefore,timely and accurate detection of pipeline leakage is one of the current research hotspots.To improve the accuracy of pipeline leak detection,this paper uses the Variational Mode Decomposition(VMD)algorithm to analyze pipeline signals,and designs a set of pipeline leak detection methods from signal preprocessing,feature extraction,and working condition recognition.The main research contents are as follows:Firstly,the principle of the VMD algorithm is studied,and then it is used to analyze the abnormal signals such as noisy signals in the signal processing process.The signal decomposition ability of VMD is compared with Empirical Mode Decomposition(EMD),and the feasibility of VMD algorithm in natural gas pipeline leak detection is verified.Secondly,the VMD algorithm needs to preset parameters before use.To determine the VMD parameters value,an improved VMD method is proposed.Sparrow search algorithm(SSA)is introduced to adaptively select the optimal combination of decomposition level K and penalty factor α.Furthermore,in order to eliminate the interference of noise on pipeline leak detection,a denoising method based on VMD and dispersion entropy is presented.Use VMD to decompose the input signal into several modal components,then calculate the dispersion entropy of each mode to distinguish the effective mode and the ineffective mode.and finally reconstruct the effective modes to get the denoising signal.The experimental results of simulation signal and actual natural gas pipeline leak signal show that the denoising effect of this proposed method is better than other denoising methods.Finally,the effective feature extraction of pipeline signals is conducive to improving the accuracy of pipeline leak detection,so a feature extraction method for pipeline signal by combining VMD with Lempel-Ziv complexity analysis is put forward.Perform denoising preprocessing on the pipeline signal,then calculate the Lempel-Ziv complexity value of denoised signal,extract the LZC feature,and use it as the input of the Support Vector Machines(SVM)for training and testing.The results show that compared with other features,the LZC feature extracted in this paper have higher classification accuracy of pipeline working condition signals.