The Application of Dynamic Network Model on Gas Pipeline Leakage Detection and Localization
Pipeline has been widely used for gas-transmission, while the accidents caused by pipeline leakage happen frequently. A new method is put forward in the paper to detect and locate the pipeline leakage, which designed a feed-forward dynamic network model for detecting and locating the pipe leakage on the base of combining the high nonlinear characteristic of the NN and time-series model which can describe the dynamic system. The train samples and test samples of the network are composed of pressure and flowrate data of all kinds of working conditions (such as different boundaries, leakage at different positions). These data are derived by both experiment results and simulated results based on the actual pipeline parameters. The network is trained by dynamic BP learning algorithm. After training (namely the network meets the training goal), if the experimental data of the pipeline are input to the network, the network will judge whether the pipeline leaks or not and locate the leakage position if leakage exists. The results of experiment verify that the network can describe the gas flow-characteristic in the pipeline, as well as detect the leakage and locate the leakage position.
pipeline leakage leakage detection localization dynamic network
Yao Zhiying Peng Guangzheng Zhou Yu
School of Information Science and Technology, Beijing Institute of Technology, Beijing 100081, China
国际会议
北戴河
英文
1067-1069
2007-06-06（万方平台首次上网日期，不代表论文的发表时间）