ARTIFICIAL NEURAL NETWORKS MODELS FOR RIVER INFLOW PREDICTION

Theory of artificial neural network (ANN) modelling is defined. The multi-layer perceptron is developed with association of a sigmoid function and applied for hydroclimatological data on mountainous catchment. Statistical accuracy measures of the network model on testing and training phases are compared. Application for the prediction values of flow in case of extreme conditions. Comparison between actual and ANN predicted values of flow.