• CALIBRATION OF ANN MODELS USING GENETIC ALGORITHMS

Abstract

In the last few decades, a lot of research is carried out in the field of Artificial Neural Network (ANN), leading to emergence of the number of training algorithms. In ANN, neurons are connected via a network of paths carrying the outputs of the one neuron as input to another neuron. The use of Genetic Algorithm (GAs) has also increased in many areas of engineering. GA search in an evolutionary way to find an optimal solution for any multi-dimensional problem. A review of the literature reveals that GAs have been applied for modeling complex processes in various branches of engineering and sciences. ANN training is possible using a variety of algorithms. In the present work, some of these algorithms are tested for their efficacy along with use of GA. The problem for this purpose is considered from domain of water resources with a focus to rainfall-runoff modeling. The comparison of modeling results shows that the tested ANN algorithms perform better than GA.