Application of Support Vector Machine on Fault Diagnosis for Water Distribution Systems

作者：Hong Zhu Xiao^*; Bao Zhuo Yan; Jun Yang; Jun Sang

来源：Disaster Advances, 2012, 5(4): 811-815.

• Support Vector Machine
• Pattern recognition
• Fault Diagnosis
• Water distribution System

For the fault diagnosis of urban water supplying pipeline network, an indoor experimental model was designed. By checking the variety of node hydraulic pressure the pattern classification of the support vector machine (SVM), which was constructed based on the theory of Structure Risk Minimization, was employed to determine the corresponding feature vectors and kernel function parameters. The samples were selected for training and testing to locate the bursting point of the water supplying pipeline network with small quantity samples. Based on some testing data for the experimental model, the SVM based fault diagnosis method for water supplying pipeline network was tested and compared with the artificial neural network (ANN) base method. The experimental results demonstrate that the fault diagnosis method based on SVM is more precise than that based on ANN.

• 出版日期2012-10
• 单位重庆大学