Pressure-driven Background Leakage Models and their Application for Leak Localization Using a Multi-population Genetic Algorithm

摘要

Abstract Model-based techniques can accurately locate the vicinity of leak localization. However, in the traditional hydraulic leakage model (THLM), the nodal flow is composed of the actual water consumption and the background leakage calculated in equal proportion, without considering the influence of pressure on the background leakage. Therefore, in this research, the parameter α is obtained through a pipe network experiment. According to the relationship between the background leakage and the pressure, the parameter β is calculated with a nonlinear genetic algorithm (NGA). The emission coefficient C is obtained based on the length of the pipeline, and then the pressure-driven background leakage model (PDBLM) is built to detect the leak localization using a multi-population genetic algorithm (MPGA). Through the simulation results under three working conditions, it is concluded that the PDBLM model is closer to the operation of the actual pipe network than the THLM model. Additionally, the PDBLM-based Inverse Problem Leak Localization Model can find the actual leakage point more accurately, improve leakage detection efficiency, and reduce water loss.