Usage of clustering methods for sequence plan optimization in steel production

Abstract

The paper deals with production scheduling of heat sequences while are steel types casted on continuous casting device. For production scheduling are used k-means clustering and fuzzy clustering methods. The parameters for cluster analysis are chemical composition liquids temperature and another values. From these values were selected parameters, which has been processed by clustering methods. Proposed clustering algorithm for sorting steel grades on continuous steel casting device is aimed to cast as many single graded smelts as possible, respectively more steel grades, which has similarities in chemical composition and liquid temperature. These resulting clusters are used when designing algorithm for smelting sequence scheduling. The goal of the production scheduling is to make schedule of production tasks, so how to achieve the agreement between order requirements and capabilities of production in given time scale.