Fuzzy Interval Oxygen Estimation in an Electric Arc Furnace from Scarce Output Measurements

In this paper, two approaches to fuzzy prediction interval modelling of the processes with scarce output measurements are presented. Many real-world processes exhibit a significant drawback, originating from infrequent and rare measurements of the critical process variables. The idea behind the presented approach is to develop a model that can estimate the unmeasured process variables and find the narrowest possible bands of these variables that contain the prescribed percentage of data, i.e., the lower and upper bounds. In this work, the fuzzy prediction interval models are applied for estimation of the dissolved oxygen content in the steel bath in an electric arc furnace. Each measurement of the dissolved oxygen content imposes an operational delay and an unnecessary loss of energy. The fuzzy prediction interval can be used in the decision-making process to reduce the number of dissolved oxygen measurements required and provide additional information to the process operators. The approaches are implemented using real operational data from the studied electric arc furnace and the effectiveness of the fuzzy prediction interval methods is illustrated.