M. Yu. Ryabchikov, B. N. Parsunkin, S. M. Andreev, N. A. Golovko

Zinc-coated rolled sheet received on the unit of continuous hot galvanizing is in great demand. One of the major objects during the galvanization process is to provide technologically preset heating modes infringement of which can lead to defects of a strip. Considering complex conditions of radiating heat exchange during heating of a moving strip in the lingering furnace of the tower type, if some considerable volume of the aprioristic information about work of the facility for mathematical model of heating is present it is the most expedient to use the model created on the basis of an artificial neural network (ANN). At the same time the considerable non-uniformity inherent in the technological data (some modes, as a rule, are used more often, than others) and dependence on time complicates complication of ANN of the model sufficient for definition of the process, requiring preliminary processing of experimental data from the developer. The data processing collected during the long term work of the unit for the purpose of creation of models of process on the basis of artificial neural networks is considered step-by-step in the article. Restrictions of a scope of the created models are specified and the suggested thermal mode control system which considers the results of modeling process is described.