Research on Gas Pipeline Leakage Detection and Localization Based on Diagonal Recurrent Neural Network

摘要

A new method is put forward to detect and locate the gas pipeline leakage, which designed a dynamic network model DRNN. The samples of the network are composed of pressure and flowrate data of all kinds of working conditions. The experiment results verify that the model can describe the gas' flow - characteristic in the pipeline, as well as detect the leakage and locate the leakage position.