Prediction of Monthly Inflow to Dam Reservoir using ANN and ANFIS: A Case Study of Latyan Dam

Precise prediction of inflow has been an important challenge for researchers during last decades. Considering the inflow prediction complexity, many prediction models have been used. Recently, ANN and ANFIS have been widely used to find the relationship between inputs and outputs without considering the physical mechanism of the phenomena. In this study, using ANN and ANFIS and based on recorded discharge, temperature, and rainfall data of Latyan hydrometric station, five different prediction models were developed to predict the monthly inflow to Latyan dam reservoir using recorded discharge, temperature, and rainfall data of Latyan hydrometric station. The best network model and its parameters were distinguished using some performance indices including CORR (Correlation coefficient), RMSE (Root Mean Square Error), and MAPE (Mean Absolute Percentage Error). The CORR, RMSE, and MAPE indices were obtained as 0.808, 5.71 and 192.58 for the best model of ANN; and as 0.825, 5.39 and 158.18, for the best model of ANFIS. Comparing the ANN and ANFIS results showed the superiority of ANFIS to ANN in modeling and predicting the inflow. Considering the results ANFIS could be used as a reliable approach for inflow prediction in the cases with high complexity and low possibility for measuring data directly.