Multi-Objective Optimization of Electric Arc Furnace Using the Non-Dominated Sorting Genetic Algorithm II

Combining classical technologies with modern intelligent algorithms, this paper introduces a new approach for the optimisation and modelling of EAF-based steel-making process based on multi-objective using evolutionary computing machine learning. Using large amount real-world historical data containing 6423 consecutive EAF heats collected from melt shop in an established steel plant work not only creates learning models both ladle furnaces but also simultaneously minimises total scrap cost energy consumption per ton scrap. In process, several algorithms are tested, tuned, evaluated compared before selecting Gradient Boosting as best option to model analysed. A similar is followed selection algorithm. For task, six techniques tested hypervolume performance indicator just then select Non-dominated Sorting Genetic Algorithm II (NSGA-II) option. Given applied research focus real manufacturing constraints variables such individual price, availability, tap additives ambient temperature used developed here. comparison equivalent literature showed 13% improvement mean absolute error usage prediction comparative metric. The resulted reductions costs that ranged 1.87% up 8.20% among different grades ranging 1.15% 5.2%. optimiser were ultimately deployed graphical user interface allowing melt-shop staff members make informed decisions while controlling operation.