Research On Intelligent Diagnosis Technology Of Gas Pipeline

China’s natural gas industry has developed rapidly.As an important part of modern transportation system,pipeline transportation bears nearly 99% of the natural gas transportation tasks in China.With the continuous expansion of the scale of the pipeline network,affected by the complexity of the system structure and the uncertainty of internal and external conditions,abnormalities or failures will inevitably occur in the production and operation process.The lighter will reduce the system performance and affect the production efficiency.This will cause leakage,explosion and other accidents,stagnating the production system,and endangering the safety of people’s lives and property.Therefore,the status monitoring and abnormal diagnosis of the gas pipeline production process are directly related to the safety and stability of the system operation,and have important practical significance.With the improvement of data acquisition level and the development of Internet of Things technology and intelligent sensor,a large amount of real-time process data is generated in pipeline system.System anomaly diagnosis based on process monitoring data is an effective means to avoid serious accidents.However,the existing pipeline monitoring system has a single way to deal with massive data,and the managers are weak in real-time analysis of production data.In addition,the flow inside the pipeline cannot be accurately quantified only by the collected data,and the running state of the pipeline network still depends on the analysis and judgment of the dispatcher.According to the above problem,based on the industrial production in the operation of gas pipeline,on the basis of the data and information on the physical mechanism of the pipeline system,analysis the characteristic,process mining combined with mathematical statistics,machine learning and computer simulation methods,such as intelligent processing on pipe production data analysis and use,and abnormal condition diagnosis technology research.The main contents of the study are as follows:(1)Using the fault tree analysis method,establish the fault tree model of the three major research objects of instrument communication,pipeline lines and station equipment,clarify the failure types and failure causes of the pipeline system,and sort out the failure parameter types and performance of the system.Finally,From the three aspects of data quality,range change and parameter combination form,the model characteristics of pipeline system failure parameters are analyzed and summarized.(2)Aiming at the univariate anomaly diagnosis of pipeline,the paper proposed to combine measurement data with control chart theory,studied six control chart modes of measurement data,obtained sample data of different modes by Monte Carlo method,and extracted the statistical and shape characteristics of sample data.As the input,four machine learning methods of neural network,support vector machine,CART decision tree and random forest were used to carry out the simulation experiment of pattern recognition.It is proved that random forest has the advantages of high precision(about 97.5%)and short time,which is more suitable for real-time status recognition of pipeline production data.A case study shows that the proposed method can provide a reliable basis for the diagnosis of univariate anomalies.(3)Aiming at the diagnosis of abnormal operation of the pipeline system,a pipeline dynamic simulation model based on the characteristic line method was established,and the Runge-Kutta method was used to realize the accurate solution of the initial value of the dynamic simulation.There is 0.53%,which can accurately describe the flow information inside the pipeline and provide reliable data for the diagnosis of abnormal system operation.Then,a diagnosis model for instrument drift,abnormal pipeline storage and abnormal gas transmission efficiency is established,and the calculation,analysis and judgment of real-time pipeline data and dynamic simulation data are integrated to realize the diagnosis of abnormal operation of the above pipeline system.(4)Based on the above research results,a gas pipeline abnormal operation diagnosis system was designed and developed.Design the overall architecture of the system,design the system database based on MySQL,establish related data tables,and implement system development based on the Intelli J IDEA platform for timely warning of abnormalities in the pipeline production and operation process,and finally display the application of the system.