Optimization of the lateral velocity distribution in rivers using Genetic Algorithm

Alluvial rivers generally have wide widths، and hence، the lateral velocity profile in this condition is very variable. Calculation of this profile is necessary in many procedures such as determination of flow discharges in main channels and floodplains، sediment transport in rivers، pollutant transport، shear stress prediction and identification of reaches with erosion or deposition across the river. Currently، the field measurements of point velocities across the river using the current meter or quasi two dimensional mathematical models are being used for lateral velocity profile prediction. Field measurement of velocity takes much cost and time and is dangerous during flood. Also، mathematical models theory is rather complex and may be make some limitations for hydraulic engineers. In this study by using genetic algorithm، a simple and dimensionless equation has been derived for velocity profile prediction. Dimensionless variables of relative depth and coherence parameter have been used as input parameters. For this study، experimental and field data of velocity distributions have been categorized to two groups of calibration and validation. With statistical comparison of results، the determination coefficient and root mean squared error of calibration data were obtained as 0. 93 and 0. 063 and for validation data as 0. 91 and 0. 127، respectively.