Use of neural networks in predicting the mechanical properties of steel

V diplomskem delu smo predstavili povezavo med dvema različnima področjema, in sicer umetnimi nevronskimi mrežami in jeklarsko industrijo. Povezavo je predstavljalo izbrano programsko orodje, NeuroIntelligence, zasnovano za delo z nevronskimi mrežami. S programom smo želeli dokazati uporabnost le-tega pri napovedovanju natezne trdnosti jekla na osnovi zbranih podatkov že izdelanih jekel. Poleg tega smo lahko praktično prikazali vpliv sestave jekla na njegovo natezno trdnost. Potrebne podatke za obdelavo, bilo jih je blizu štiri tisoč, smo pridobili v podjetju Metal Ravne d.o.o., ki se ukvarja s proizvodnjo jekla.In this diploma work, we have presented a connection between two different areas, artificial neural networks and steel industry. Connection between those two areas is selected software, NeuroIntelligence, which is designed to work with neural networks. With this program we demonstrated the prediction of tensile strength of steel, based on a data that we collected from ready-made steel. We may also practical present the influence of the composition of the steel to its tensile strength. Necessary data information for treatment, there were close to four thousand, we acquired from the steel company Metal Ravne d.o.o.