Leakage prediction of water supply network based on SAA-SVM model

Abstract

In order to explore the relationship between leakage, leakage failure form and pipeline pressure, the SAA-SVM model is established based on Support Vector Mechanism. The simulated annealing algorithm is used to optimize the parameters c and g of SVM, which greatly improves the accuracy of SVM and the ability to jump out of the local optimal solution. In laboratory, the leakage data under various working conditions are measured through the water supply network model. It is input into SAA-SVM model as sample data set for optimizing the model by machine learning. The results show that the optimized SAA-SVM can predict the leakage shape and loss quickly and accurately.