Predikce rychlosti atmosférické koroze a korozní agresivity pomocí neuronových sítí

Abstract

This thesis is focused on the usage of the artificial neuronal network through which the speed of an atmospheric corrosion is predicted. Theoretical analysis of the atmospherical corrosion and the artificial neuronal network is mentioned in the first part. The aim of the thesis is to design and verify the artificial neuronal network model. The model will predict corrosion loss of the structural carbon steel according to enter environment variables. Second part includes the creation of the predictor artificial neuronal network model which is constructed by Statistica – Neuronal network program. At the close of the thesis the achieved results are compared and summed up, including the description of practical application.