Study on Precise Temperature Control Method of Hot-rolled Ribbed Bar during Cooling Process

Abstract

Aiming at the demand for performance control in the production process of hot-rolled ribbed bar, the finite difference method was used to establish a temperature prediction model during the cooling process. The calculation accuracy of the temperature model depends to a large extent on the selection of heat transfer coefficient. In this study, the industrial production data was cleaned and screened by the clustering algorithm to obtain training sample data, and based on BP neural networks, the mapping relationship between different influencing factors and heat transfer coefficient was established. According to the change of production conditions, the heat transfer coefficient is learned adaptively, which improves the prediction accuracy of the model. By comparing the predicted and actual temperature values over a period of time, the deviation between most predicted and actual values is less than 20°C, and the deviation is reduced by about 30°C compared with that before the BP network was used to study the heat transfer coefficient. Under the constraints of the target temperature and the temperature difference between the inside and outside of the section, the precise control of the cooling temperature of the hot-rolled ribbed bar is realized by the calculation of the temperature prediction model.