Automated generation of reference steel-melting profiles in an electric arc furnace

Svetovne potrebe po jeklu so začele, z vstopom v 21. stoletje, strmo naraščati, kar je privedlo do porasta cen surovin. Jeklarne stremijo k optimizaciji procesa taljenja jekla in rešitve iščejo v modernejših konfiguracijah talilnih peči z boljšimi izkoristki. Poleg gradnje novih izboljšanih elektro obločnih peči se jeklarska industrija posveča tudi optimizaciji delovanja obstoječih sistemov. Delo predstavlja postopek razvoja orodja za avtomatsko generiranje optimiziranih profilov taljenja jekla v elektro obločni peči. V prvem delu se posvetimo spoznavanju z elektro obločno pečjo, njenimi podsistemi in pojavi znotraj te. Ta del je ključen za razumevanje problema, ki ga poskuša delo reševati. Drugi del zajema analizo podatkovne baze o obratovanju elektro obločne peči, z namenom odkrivanja vhodno-izhodnih relacij procesa in pridobivanja znanj o lastnostih posameznih spremenljivk v podatkovni bazi. Predlagamo model, ki ob znanem zalogu elektro obločne peči predlaga nove profile taljenja. Relacije med zalogi smo pojasnjevali z različnimi merami podobnosti. V zadnjem delu magistrske naloge predstavimo grafični vmesnik, ki združuje funkcionalnosti pregledovanja podatkovne baze, preverjanja delovanja različnih mer podobnosti med zalogi in avtomatskega generiranja profilov taljenja jekla.In the beginning of the 21st century, world steel needs began to rise steeply, which led to a rise in raw material prices. Steelmakers strive to optimize the steel melting process and are looking for solutions in more modern configurations of melting furnaces with better efficiency. In addition to the construction of new improved electric arc furnaces, the steel industry is aiming to optimize the operation of existing systems. The thesis presents the procedure of developing a tool for automatic generation of optimized steel melting profiles in an electric arc furnace. In the first part, we focus on presenting the electric arc furnace, its subsystems and the phenomena occurring inside the furnace. This part is crucial to understanding the problem, the work is trying to solve. The second part covers the analysis of the database, containing operational data of the electric arc furnace, in order to detect the input-output relations of the process and to acquire knowledge about the properties of individual variables in the database. We propose a model that generates new melting profiles in relation to the specified charging of an electric arc furnace. The relations between the charges were explained by different similarity measures. In the last part of the thesis we present a graphical user interface that combines the functionality of exploring the database, examining the behavior of various similarity measures between charges, and automatic generation of steel melting profiles