Blast Furnace Cross Temperature Prediction Based on Data-driven and Intelligent Optimization

Least square support vector machine (LS-SVM) is employed to model of the blast furnace cross temperature. First,correlation analysis is performed to select inputs related to cross temperature. Second,an improved particle swarm optimization (PSO) is proposed to obtain optimized parameters for LS-SVM in order to improve the prediction accuracy. Finally,the prediction model of blast furnace cross temperature based on LS-SVM is achieved. Experiments using practical production data illustrate that the input selection based on correlation analysis can reduce the computation complexity without influencing the prediction accuracy. Compared with the grid search method,the proposed data-driven cross temperature model represents an improvement in accuracy of 3%,and can meet the requirements of production.