Temperature Prediction Model of Cement Rotary Kiln Based on MPGA-LSSVM

Abstract

Abstract. The burning zone temperature (BZT) in cement rotary kiln is a mostly important index which affects the quality of produced cement. In rotary kiln, the bilateral movements of air and complex chemical reactions take place. Thus, traditional modeling methods can not predict the BZT effectively. In order to improve the prediction accuracy, the temperature prediction model was established by least square support vector machine (LSSVM). The input variables of this model were selected by the gray correlation analysis method and mechanism analysis. The parameters of the LSSVM were optimized by multi population genetic algorithm (MPGA). The final simulation results show that the prediction accuracy is improved obviously. The effectiveness of the proposed method is verified.