A New Data Preprocessing Technique Based on Feature Extraction and Clustering for Complex Discrete Temperature Data

Abstract

LF(LADLE FURNACE) refining technology is the key process to regulate the temperature in steelmaking process. To predict the end temperature of molten steel in LF, this paper proposes a new data preprocessing technique based on feature extraction and clustering. Firstly, random forest algorithm was used to predict the temperature, the predictive hit rate of error within ± 10°C was 73.18%. The Lasso algorithm and K-means algorithm was used for feature extraction and clustering. After improvement, the prediction accuracy of the LF end temperature of error within ± 10°C was about 88.16%. The results show that this improvement has high prediction accuracy in the prediction about the end temperature of molten steel in LF refining.