Research On Flatness Prediction Control In Rolling Process

Plate and strip are widely used in various fields of our national economy. Higher and higher requirements are put forward on the quantity and quality of plate strip. In order to improve the quality and value of the plate and strip products, pursuing the good profile quality should gets the priority in an iron and steel enterprise. Therefore, researching on the control of the plate and strip is a meaningful project.This paper took the4-high plate and strip cold rolling mill as the object of study and for the purpose of precise online control of the plate crown, further study on the influential factors of the plate crown and establishment of an accurate prediction model of the plate crown have accomplished. The main contents are as follows:(1) Finite element rolling model of four roller cold rolling mill is established by using ANSYS/LS-DYNA. The change of flatness in the strip steel rolling process is emulational analysed. Then the finite element model is solved and analysed, including the analysis of overall simulational results, the plate and strip simulational results and the roll simulation results in the plate and strip rolling process. Based on the finite element simulation, the influence of the plate crown on the roll bending force, roll diameter, roll crown, roll shifting quantity and strip width is studied selectively, and the change regulation of plate crown is analysed by using MATLAB.(2) The forecast accuracy of the plate crown parameter in the flatness prediction model is improved by introducing BP neural network. Based on the concept of artificial neural network, the learning algorithm, the application situation in the flatness control and the optimization algorithm, the construction process of BP neural network model is introduced in detail, including the selection of input and output parameters, the structure determination of the neural network model.(3) According to the Finite element analysis and BP neural network model, the overall flatness prediction model of the combination between the Finite element and the BP neural network is established. Using the results of finite element simulation analysis as training samples and testing samples for training and testing BP neural network, the prediction precision of the flatness prediction model of the combination between the Finite element and BP neural network and the traditional shape prediction model is compared. Results show that the prediction accuracy of the flatness prediction model of the combination between the Finite element and BP neural network is much higher than that of traditional model.(4) The experimental study was carried out with the2800four roller cold rolling mill as the experimental platform, further studied the influence of the plate crown on the roll bending force. Comparing the experimental values with the calculation values of the finite element model, through comparative analysis, it obtains that the change regulation is consistent and verifys the correctness of the finite element model.In conclusion, this paper established the flatness prediction model based on finite element combined with BP neural network, and analysed key factors that influence the plate crown, and predicted the plate crown parameters, expecting to provide the certain theoretical foundation in improving the accuracy of flatness control.