Image Processing & Communication, vol. 17, no. 4, pp. 307-312 DOI: 10.2478/v10248-012-0059-2 307 COMPARISON OF METHODS FOR DETERMINING THE THERMAL CONDUCTIVITY IN INDUCTION HEATED INDUSTRIAL ROTATING CALENDERS PIOTR URBANEK,JACEK KUCHARSKI ,ANDRZEJ FR˛ ACZYK Institute of Applied Computer Science, Lodz University of Technology, Poland piotr.urbanek@p.lodz.pl Abstract. In many technical applications it is necessary to know thermal diffusity and heat conduction in solids, liquids and gaseous. In case of solid bodies the experimental determi- nation of those properties requires use of special laboratory stands which provides the appropri- ate initial and boundary conditions necessary to solve invert problems. In the paper two meth- ods of determine thermal property of metals has been presented and discussed. First method is based on classical optimization methods and the second one by artificial neural network, which is trained with data from numerical model of in- vestigated body. Both method were tested on real laboratory model. 1 Introduction In many technical applications it is necessary to know ma- terial properties e.g. thermal diffusity or heat conduction in solids, liquids and gaseous. Methods of experimen- tal determination thermal properties of solid bodies needs the special laboratory stands which provides the appro- priate thermal initial and boundary conditions. Generally, it is important to know how much energy is supplied to test body, and what is the temperature gradient between two different characteristic places of examined body. In the paper two methods for determining thermal property of moving cylinder have been presented. First method is based on classical optimization and the second one by ar- tificial intelligence (artificial neural network). Both meth- ods were tested on data of temperature distribution along calender axis recorded by infra-red camera. 2 State of the art of methods of determine thermal properties of solid bodies Widely used classical methods are based on ensuring the special heat exchange conditions. The schema of experi- mental stand is shown in Fig. 1. Using results of the experiment (Fig. 1(a)) the thermal conductivity of sample could be determined by the opti- mization method (shown in Fig. 1(b)). Using, for instance, the finite difference method heat flux density can be calculated from second boundary con-