Predicting foaming slag quality in electric arc furnace using power quality indices and ANFIS

Abstract

Foaming slag quality is an important parameter that can be used to improve the efficiency and quality of electric arc furnace process. However due to its fast and unpredictable changes, its quality is difficult to control. In this paper, an Adaptive Neuro Fuzzy Inference System (ANFIS) is used to determine slag quality based on power quality indices in electric arc furnaces. In order to train the intelligent system, a power quality analyzer is installed on an electric arc furnace feeder to record its power quality parameters. All electrical power quality parameters have been measured for 13 melting. Twelve groups of power quality parameters are examined for prediction slag quality and finally one group including total current harmonic distortion, seventh current harmonic, and three phase current unbalance are selected which shows the best prediction accuracy. The intelligent system trained by six melting data and tested experimentally by connecting power quality analyzer to furnace feeder to predict the slag quality every minute. Experimental results show the accuracy of prediction is about 95%. The designed intelligent system can also be used in slag control process.