Research On Gas Pipeline Leakage Diagnosis Based On Resistance Coefficient Identification

With the constantly growing number of natural gas users,the natural gas pipe network is becoming more and more large and complex,which brings the convenience to the whole society,but also a certain security threat was brought thereupon.For the gas pipe network,the traditional management mode has been unable to meet the existing security requirements.Based on the rapid development of information technology,it has become an efficient new management method to monitor the running status of pipe network by using intelligent algorithm and pipe network simulation.On the basis of this,this thesis establishes the pipe network leakage diagnosis model through genetic algorithm and BP(Back Propagation)neural network algorithm.The pipe network leakage diagnosis model handles the operation data of the pipeline network through genetic algorithm,recognizes the resistance coefficient of the pipe network,realizes the optimization of the pipe network simulation system,and reduces the difference between the simulation results and the real situations.After the optimization of the simulation system,it simulates the leakage in different situations of the pipe network,processes the simulated leakage data by using BP neural network algorithm,and obtains the mapping between the leakage position and the pipe network pressure by practice,which enables the diagnosis model to fix the position of leakage through the pressure change of the pipe network,providing technical support for the online leakage monitoring of the gas pipe network.Meanwhile,this thesis also studies the influence of the pressure monitoring nodes layout on the leakage diagnosis model of pipe network,puts forward the adjustment scheme for the models with insufficient number of monitoring nodes,and developes the application software of pipe leakage diagnosis based on MATLAB /appdesigner.This thesis adopts the literature data to verify the established pipe network simulation system,and the error rate of hydraulic computation results is basically maintained at around 1% with scientific and reliable system.The resistance coefficient identification of a medium-pressure annular pipe network is verified,and compared with the real cases,the maximum relative error of the node pressure with the calculation of the common resistance coefficient is 89.3%,and the maximum relative error is less than 10% after the resistance coefficient identification,which greatly improves the simulation accuracy of the pipe network.The leakage of a high-pressure tree-structured pipe network is diagnosed,and the correct rate of judgment on the leakage pipe number is more than 95%,and the error rate of judgment on the specific location of leakage point is less than 5%,which can satisfy the requirements of engineering practice.The software developed by this thesis has the functions of hydraulic calculation,resistance coefficient identification,leakage working condition simulation,leakage diagnosis and monitoring nodes layout optimization with simple and clean user interface,friendly operation,and intuitive calculation results,which is of a certain promotion value.