Investigating Application of Adaptive Neuro Fuzzy Inference Systems Method and Epanet Software for Modeling Green Space Water Distribution Network

Abstract

By the effect of pressure or velocity fluctuations, the water supply networks may be damaged. For avoiding of this case, suitable and optimized management of networks is very necessary. In this research, control of pressure and velocity was investigated to prevent problems of water supply network and also hydraulic characteristics were also predicted by Adaptive Neuro Fuzzy Inference System (ANFIS). In this way, first by zoning the city of Kangavar province (as case study) to six sub-zones based on distribution parameters, plotting of water supply networks with 10 years design period and target population of 95,000 based on 22 h in day irrigation and 29.6 square meter per capita at the end of design period were used. Pressure and velocity of network were then analyzed utilizing EPANET software. Based on the results, maximum pressure occurred at 3–3 node in third pressure region which estimate 100 m of water and maximum value of network velocity was estimated as 1.4 m per second. Also, results showed that the discharge was used in model according to diameter of tubes and selected paths in different regions are in appropriate range. Then, using the measured values, the fuzzy neural network was trained with the help of particle swarm algorithm, genetic algorithm, differential algorithm and ant colony algorithm optimized, and the optimal network for velocity prediction was obtained from fuzzy neural network with ant colony algorithm while for head loss fuzzy neural network with particle swarm algorithm was selected as the best model. According to the sensitivity analysis, diameter of pipes was found to be the most effective parameter in predicting velocity and pressure loss. Results showed the high capability of ANFIS in analyzing and predicting hydraulic properties of water supply tube.