Simulation of Rainfall-Runoff Process using multilayer perceptron and Adaptive Neuro-Fuzzy Interface System and multiple regression (Case Study: Khorramabd Watershed)

The discharge or runoff which ousts from a watershed is important. Because its deficiency leads to financial losses and its excesses cause damage in lives and property as flood. In this research using Artificial Neural Network Multi-layer Perceptron (MLP (and Adaptive Neuro-fuzzy interface system (ANFIS) and multiple regression method simulated rainfall- runoff process on daily basis in the Khorramabad watershed. For inputs, different combinations of precipitation inputs including current rainfall, pervious day rainfall and two previous days were used. Inputs membership function for ANFIS model in this research is: the trapezoid, triangular, Gaussian and Gaussian type 2. MLP model that used in this research, was evaluated with one hidden layer and the number of variables neurons. The results showed that Adaptive Neuro-fuzzy interface system (ANFIS) compared to multi-layer perceptron model (MLP) and multiple regression model, has better performance. Also by increasing in the number of inputs, involvement pervious day rainfall and two previous days, all three models performance will be better.