Natural Gas Pipeline Leakage Detection Based On Digital Twin

With the rapid development of our technology and economic development,various large-scale enterprises,people’s demand for energy is also increasing.Some energy industries such as oil and natural gas also begin to emerge and grow.Pipeline transportation not only has a large amount of transportation,low cost,but also can realize automatic control,has become the main mode of land energy transportation in our country.With the continuous expansion of pipeline network,a variety of pipeline safety problems are also exposed.Corrosion aging caused by long-term operation of pipelines and the theft of natural gas by criminals,pipeline leakage occurs from time to time.Once pipeline leakage occurs,it will cause serious economic losses and safety problems.Therefore,it is urgent to develop effective pipeline leakage detection methods.In this paper,a set of leakage detection method of natural gas pipeline based on digital twin is designed.The main research contents are as follows:First,a visual model of the physical pipeline is constructed.The construction of visual model is the key technology to build pipeline digital twin,which not only needs the mechanism model of physical object,but also needs the ability of real-time simulation and three-dimensional visualization effect.In this paper,through the finite element method,Ansys software is used for 3D modeling and simulation of the physical pipeline,and the visualization model of the pipeline is established through the 3D ROM plug-in.Secondly,the leakage identification model of pipeline is constructed.By using the constructed visual model and the collected pipeline pressure data,the feature Vector was built and input into the Support Vector Machines(SVM)for condition recognition.The recognition accuracy was 86%.The locust algorithm was used to optimize the SVM parameters c and g.The recognition accuracy of SVM is improved to 90.5%.Then,the collected pipeline data contains a large amount of noise,which will inevitably limit the recognition accuracy of SVM,so an improved Variational Mode Decomposition algorithm(VMD)has been studied: Aiming at the problem that it is difficult for VMD algorithm to select preset scale K and distinguish effective mode from noise mode after decomposition,a new method was proposed to determine preset scale K and Kendall Tau correlation coefficient(Kendall Tau,KT)a joint criterion method for determining effective modes and applying it to pipeline data denoising.Compared with other denoising algorithms,the improved VMD algorithm has better denoising effect.Finally,the digital Twin of gas pipeline leak detection is built by using Twin Builder software.Among them,the pipeline data update model,leak identification model and visualization model are constructed.The pipeline data update model constructed uses sensor to collect data and upload it to My SQL database,and uses Java script to read the pipeline data and de-noise it before passing it into Twin Builder platform.Twin Builder input the pipeline data and visual model output data into the leak identification model for identification,and the identification accuracy reached 96.5%.The leak identification model returned the identification results to the visual model for three-dimensional display.The constructed digital twin realizes the visual detection of pipeline operating conditions.