Aimed at the position system of leakage identification of water supply pipeline, based on the time series data witch are formed by pipe network pressure and flow parameters, the paper applied support vector machine method to set up leakage identification model. By using particle swarm algorithm to optimize the parameter c, g in support vector machine (SVM) ,it eventually realized accurate positioning through using the method of pressure gradient. The experiment shows that the accuracy of established model of pipeline leakage positioning is higher and can meet the requirements of leakage monitoring of water supply pipe network.