Author(s)

M. Pohanka, M. Raudensky & J Horsky

Abstract

The paper describes a method of improving parameters of an experimental data based mathematical model of cooling by water-nozzles using a neural network. Experiments are carried out using hot flat and cylindrical moving objects. These experiments simulate industrial applications such as cooling in continuous casting, cooling of products in hot rolling, and cooling of rolls in rolling technology. The measurements are evaluated using the inverse 3D task. The inverse task computes the surface temperature history and heat transfer coefficients. Using the heat transfer coefficients, the parameters for the mathematical model of cooling are obtained for a specific nozzle. Their correctness is verified using the computer simulation of

Keywords