The use of optimized artificial neural network model by the Genetic Algorithm in estimating water salinity parameters (Case study: Gorganrood River)

Salinity and the amount of existing Sodium are one of the most important qualitative criteria in water classifying in terms of agriculture which makes it necessary to manage and plan this sector due to lack of fresh water and high volume of saline lands in the country. In this research, the use of intelligent Genetic Algorithm (GA) are consideredto improve artificial Neural Network performance (ANN) for estimating qualitative parameters of Gorganrood River such as Electrical-conductivity (EC) and Sodium adsorption ratio (SAR). Therefore, at first by sensitivity analysis, the input parameters for each system including The ANN and ANN-GA were determined and then were used to estimate each of qualitative parameters. The evaluations indicated that for estimating ES and SAR parameters, ANN-GA model with R2=0.95, RMSE=14.7 and MAPE=7.65 for EC and R2=0.89, RMSE=0.57 and MAPE=11.4 for SAR showed better results than ANN model with R2=0.89, RMSE=380 and MAPE=9.6 for EC and R2=0.65, RMSE=1.15 and MAPE=13.85 for SAR.

کلیدواژه ها:

Gorganrood River ، Genetic Algorithm ، Sensitivity analysis ، Artificial Neural Network ، Water Salinity