以某厂热连轧精轧设定模块为研究对象,在其原有设定模型中添加了另外一种变形抗力模型,通过计算机编写程序,创建回归系数文件,修改输入计算模块,用新添加变形抗力模型回归过的钢种系数对现场生产中与其相近的钢种进行设定计算,比较变形抗力模型添加前后的轧制力预测值偏差。结果表明:添加变形抗力模型后的轧制力计算可以得到比较满意的精度;平滑系数在自适应学习中是十分关键的影响因素,在总结了精轧机组自适应模块中平滑系数的影响因素后,建立平滑系数的动态优化模型,通过编程读取实测数据计算轧制力,比较优化前后的轧制力误差,结果表明对于精轧机组自适应过程,动态优化平滑系数更有助于提高轧制力模型预报精度。
The calculation process and file composition of the pre-setup model for finishing rolling of hot continuous rolling, were investigated. Another deformation resistance model was added to its original model by creating regression coefficient file and modifying input and calculation model using C language. Compared with the predicted values of rolling force with original resistance model, the result demonstrates that calculation precision of rolling force with modified resistance model could be satisfactory. Smoothing coefficient was a key factor for self-adaptive. After the influence factors of smoothing coefficient were summarized, the dynamical optimized model of smoothing coefficient was established. Then the rolling forces were calculated using the programming with actual data using the dynamical optimized model. The analysis of the errors after optimization clearly shows that dynamical optimization of smoothing coefficient is more helpful to improve the prediction precision of rolling force model.