تخمين دبي متوسط ماهانه با استفاده از شبكه هاي عصبي مصنوعي مطالعه موردي: آبخيز قشلاق سنندج

Estimation of the average monthly discharge using Artificial Neural Network Case stady: the Queshlaqʹs watershed of Sanandaj

Precise prediction of monthly average discharge values input to water resources such as dams has a basic role in their planning, management, sustainable and optimal operation. Given the input discharge value to dam, the annual input water volume can be calculated, and well-management for water optimum allocating to various consumption sectors, including edible, agricultural, hydro-electrical production can be scheduled. There are various parameters aflfccting the input discharge value. They are not fully known, and their relationship with input discharge is non-linear and complex. Thus, giving analytical and mathematical relationship of this concern is difficult and impossible. Artificial Neural Networks, due to their unique properties, have high abilities in non-linear and complex relation simulation. In this study it is attempted to design multi-layer Perceptron with Back Propagation learning rule for recovering the non-linear relationship between dependant and independent variables, so that, using it, prediction of monthly average input discharge to Queshlaq dam could be done. For further validation of the proposed model, obtained results from neural network model were compared with the ones obtaining from Khoslaʹs empirical method. The results from the study showed that there is an acceptable overlapping between predicted values from Artificial Neural Networks and observed data, as well as the proposed neural network model and Khoslaʹs empirical method predicts the monthly average discharge with root mean square error as 1.49 and 11.88 respectively.