Calibración de parámetros en redes de riego mediante algoritmos genéticos

Abstract

This paper presents a calibration methodology for adjusting the parameters of an irrigation network. The objective is to adjust the hydraulic model to reflect more accurately the real operating conditions. For this purpose, data from pressure and flow sensors are used. The decision variables are pipe roughness, which plays a key role because it affects distribution head losses. This problem is addressed by using genetic algorithms, since the number of decision variables is very large and the hydraulic model is nonlinear. The work analyzes the impact on the calibration results of considering roughness by material type or by independent pipe. The methodology is applied to a real irrigation network of a 125ha citrus farm in Huelva, showing an improvement in the prediction capacity of the calibrated model of 10.8% in the Willmott index for pressures, compared to the initial model designed with the producer’s data.