Tvorba odtokového modelu pomocí neuronové sítě

Abstract

This diploma thesis deals with creation of rainfall-runoff model using neural network. The theoretical part is focused on research work in the field of hydrological modelling, including rainfall-runoff process problematics and models which are used. Another topic of this research are neural networks, and in the end also the using of neural networks in hydrology. In the practical part there are created models of different schemas (input data and the count of hidden layers) using software RStudio with package neuralnet. These models simulate relationships between precipitation and flow on the river Bělá in Jeseníky. Based on models there were also made testing predictions for comparison between networks and deciding which network makes the best results.