Data-Driven Modelling and Optimization of Energy Consumption in EAF

In the steel industry, optimization of production processes has become increasingly important in recent years. Large amounts historical data and various machine learning methods can be used to reduce energy consumption increase overall time efficiency. Using from more than two thousand electric arc furnace (EAF) batches produced SIJ Acroni steelworks, electrical during melting was analysed. Information on consumed each step process is essential efficiency EAF. paper, four different modelling approaches for predicting EAF operation are presented: linear regression, k-NN modelling, evolving conventional fuzzy modelling. phase, a set ten regressors, only those that have greatest impact were selected. The obtained models accurately predict determine optimal duration transformer profile melting. by selecting pre-processed training data, where main steps find remove outlier with highest identify influencing variables contribute most increased consumption. It should emphasised too high because unnecessarily prolonged. proposed models, operators obtain information estimated before batch processing depending scrap weight basket added additives, as well given batch. All validated compared using 30% all model particular providing accurate prediction results. expected use developed will lead reduction an