Uplatnění analytické diagnostiky při podpoře prediktivního řízení údržby kruhových krystalizátorů

Abstract

The aim of the thesis is multidimensional linear ring mold´s wearing model, which is based on fuzzy cluster analysis of casted steel grades. The model should serve as a base to support the predictive mold maintenance. According to number of smelting casts and planned smelting casts, divided into clusters of related structures, it is possible to determine the lower average taper of the mold. The lower average taper of mold is the main function parameter of diagnostic predictive maintenance. The limit value of this parameter is value “1”, or value at which the normal quality of continuously casted blanks are supposed for normal technological conditions. Created model can substitute the wearing measurement of the ring mold. For this purpose is usually used measuring system MKL 100/420. This allows the increasing availability of continuous steel casting device, with knowledge of actual and predicted condition of mold´s wearing. The other usage of model using is the support of production scheduling for continuous steel casting device, when model is going to be a base of algorithm for determination of melt´s numbers in each clusters. The combination of steel melts numbers is the result and user can choose the suitable combination according to production order. The solution is based on analytical processing of operating data according to mold wear caused by amount of casted steel grades. It´s assumed that the advantage of proposal solution is its universality for all molds of the same format, which are related to specific equipment for continuous steel casting devices. The solution rely on operational and diagnostic data from at least one mold, on which basis could be able to determine significant coefficients of each cluster concerning mold wear.