Roller Parameters Prediction of Steel Tube Based on Principal Component Analysis and BP Neural Network

In the roll position adjustment of current steel tube production line, roller parameters decision for new tube is only based on the artificial experience and field measurement. Lack of effective prediction method leads to the low yield of production line and the waste of labor cost. This paper takes the F400 steel tube production line as the research object. Firstly, the stream of variation model for roll forming process is established and used to predicting the roller parameter. Then, according to historical roller parameters, the machine learning prediction model is built with principal component analysis and BP neural network. Finally, the predictive simulation and production test based on the PCA - BP model is performed. The simulation and experiment results show that PCA-BP intelligent prediction outperforms the stream of variation model and artificial roller adjustment, and the production error can be controlled within 2mm. PCA-BP prediction method shorten the changing time, improve the adjustment accuracy of new steel tube, solve the problems of production yield and labor cost, and is advantageous to the actual production application.