Research on Leak Detection of Water Pipeline Base on PSO-SVM

Abstract

Leakage of water supply pipeline will cause waste of water resources and economic losses, the traditional support vector machine (SVM) had two free parameters (include kernel parameter g and penalty factor C from the radial basis function kernel) which could lead the classifier in an instability situation. In this paper, a method was presented to establish a SVM model which used the particle swarm optimization algorithm (PSO) to determine the two unknown parameters. Particle swarm optimization is an intelligent computational swarm optimization method, which optimizes a problem by iteratively trying to improve a candidate solution with regard to a given measure of quality. The method not only has strong global search capability, but also easily to implement. In order to improve the accuracy of the SVM model, the PSO algorithm is used to optimize the parameters of SVM. A prediction model of water pipeline leak detection based on PSO-SVM is established after the sample set is been collected, trained, tested and classified. The experimental results showed that with the strong global search capability of PSO, the free parameters of SVM was been set effectively and accuracy of the classifier had been promoted.