Hybrid neuro-fuzzy approach for flood prediction and Dam gate control

Abstract

In recent years, research on Artificial Neural Networks (ANN) prove that it offers several advantages over conventional Computing methods. ANN proves to be most convenient and easy tool for modeling and analysis of non-linear events. This ability of ANN to model non-linear events is important in hydrology to model various hydrological events which are dominantly nonlinear in nature. ANN is important, where the time required to generate solution is critical. It is also capable of Modelling Nonlinear relationship as compared to other Mathematical modeling techniques which model linearity only. In this paper ANN technique along with Fuzzy technique and their interfacing is over-viewed. This paper has two parts-Modelling of Runoff using ANN and Controlling Dam Gates using Fuzzy logic. This ANN modeling is done with the help of MATLAB neural Network tool Box. Back Propagation Algorithm is used to optimize network connection weights. Back Propagation (BP) algorithm evaluates error and back propagates it for more accurate training of ANN. Five river discharge data as inputs are considered and Runoff data as ANN output. Fuzzy Inference System (FIS) simulation is used to automate opening and closing of Dam gates. Predicted Runoff and Water level of Dam is applied as inputs to FIS to control Dam gates.