A Model for Billet Temperature Prediction of Heating-furnace Based on Improved PCR Method

This paper establishes a pivot element predictive regression model between billet temperature variable and process variables with multi-statistic projection principle and PCR method, and parameters of the model are reckoned based on the actual data from a steel works. Check and error analysis indicate that this model can predict billet exit temperature 10~25 minutes in advance, and the predicting error can satisfy the demands of industrial application accuracy.