Leakage in the Water Supply Network using EPANET Software and Neurofuzzy Method

Water supply networks as a hydraulic system of water transmission and distribution have always been of interest to researchers. The main cause of water transfer in water distribution networks is the pressure head difference between the two points and in case of increasing the standard pressure, the undesirable phenomenon of leakage occurs. Leakage in the irrigation system is economically, socially and environmentally significant. Therefore, leak detection as one of the duties of water and sewage companies in the country has always been a concern. Objectives of the present research includes the detection of water loss points in the water distribution network using the EPANET model and the neurofuzzy method, and compares the two methods and provides a better method for detecting water loss points. In this study, in order to detect leakage in the water distribution network of Mohyabad city of Kerman province, a method based on hydraulic modeling and inverse solution of flow equations to predict the location and amount of leakage in the distribution network Water was introduced using EPANET software and neurofuzzy method with measured values of pressure in a number of network nodes. The results indicate that in the best selected architecture among all tested networks, leakage coefficients resulting from neurofuzzy prediction compared to EPANET modeled values are the most suitable option for modeling and have a coefficient. The correlation between 0.984 and the mean MSE forecast error, equal to zero with the root mean square error (RMSE), 0.0024 (l/s) indicates the optimal forecast accuracy and high reliability of the trained network. The proposed method with a minimum of hydraulic information from pressures, has a good ability to predict the location of leaks in the network and also the use of neurophase method is simple and low cost, in addition, has good accuracy.