The Crown Predictive Model of Cold-Rolled Zirconium Alloy Strip Sheet Based on Machine Learning Algorithm

Abstract To solve the problem that the crown accuracy directly affects the profile and flatness of the zirconium alloy strip caused by the multi-schedule and more than sixty passes cold rolling of nuclear zirconium alloy strip. The crown predictive model is established based on support vector regress algorithm (SVR), and the parameters of the SVR algorithm are optimized by the SAPSO algorithm. Meanwhile, the crown predictive model based on the SAPSO-SVR algorithm shows that the relative coefficient of predicted values is higher than 0.94. Industry test demonstrate that the crown predictive model provides a new method for zirconium alloy strip shape control and an approach to optimize the control strategy. The proposed model provides a new method and idea for shape control and optimization research in the zirconium alloy plane rolling process.