Research on Leakage Fault Location Detection Technology of Urban Natural Gas Pipeline Network

Xiaohong Zhu, Liangji Jiang, Junyu Chen

Abstract


For the fault diagnosis of urban natural gas pipeline network, an indoor experimental model of natural gas pipeline network was designed. By periodically testing the gas pressure flow variety of the pipe network node, and using the support vector machine (SVM) classification function which based on the minimum structure risk principle, the corresponding feature vector and kernel function parameters are determined and the samples are selected for training and testing in small quantity samples, in the case of the pipe network failure point to judge. Based on the testing data of multiple experimental models, the paper tests the fault diagnosis method of natural gas pipeline leakage which based on SVM, and compares with neural network (ANN) method under the same experimental data. The actual example shows that the diagnostic accuracy of the SVM method is better than the neural network method under the same experimental test data.

Keywords


Machine Learning; Support Vector Machine; Leakage Fault Diagnosis; Urban Natural Gas Pipeline Network


DOI
10.12783/dtcse/iciti2018/29077

Full Text:

PDF

Refbacks

  • There are currently no refbacks.