In the production process of hot rolling,the control accuracy of coiling temperature is one of the key parameters that de-termine the quality of the product.Based on the heat transfer mechanism model and comprehensive analysis of actual production data and process parameters,taking into account factors such as finlshing rolling temperature and strip thickness,it is deeply studied the impact of key factors such as strip speed,cooling water temperature,and seasonal variations on the coiling tempera-ture model,and the model is revised and optimized.At the same time,the compensation model based on alloy composition was constructed using machine learning algorithms,and comparative analysis was conducted on different algorithms.The research re-sults show that the random forest prediction model performs well in improving the control accuracy of coiling temperature.The research results were applied to actual production,resulting in an increase in the average qualification rate of coiling temperature for strip with thickness h≤6.0 mm,6.0 mm＜h≤13.0 mm,and h＞13.0 mm by 3.07％,3.82％,and 4.68％respectively,providing a new and effective way to further improve the control accuracy of coiling temperature.
hot rolled stripcoiling temperatureheat transfer mechanism modeldata-drivenmachine learning algorithmscoi-ling temperature model
维普
万方数据
