Paper Info
Title
Decoupling Predictive Control of Strip Flatness and Thickness of Tandem Cold Rolling Mills Based on Convolutional Neural Network.
Abstract
The control system regarding flatness and thickness of the strip in tandem cold rolling mill is generally multivariate, non-linear, and strong-coupling. In this paper, a decoupling predictive control method based on convolutional neural network is invented for the decoupling of these two elements. We at first introduce the mathematical model, namely AFC-AGC model, of the automatic control on flatness and thickness, and analyze the coupling problem between them. Then CNN with strong feature extraction ability is used to design the decoupler of the flatness and thickness system by iteratively learning of the input and output data of the control system in consideration. Next, according to the non-linearity and the pure delay of the rolling mill system, we design the flatness and thickness controller using model predictive control algorithm (MPC). Finally, the simulations and comparisons are performed on the actual parameters of a tandem cold rolling mill. Our results demonstrate the effectiveness of the proposed decoupling control method and confirm the accuracy and robustness of the modified flatness and thickness control system.
Year
DOI
Venue
2020
10.1109/ACCESS.2019.2962544
IEEE ACCESS
Keywords
Field
DocType
Tandem cold rolling mills,flatness and thickness of strip,decoupler,convolutional neural network,model predictive control (MPC)
Flatness (systems theory),Control theory,Computer science,Control theory,Decoupling (cosmology),Model predictive control,Automatic control,Robustness (computer science),Input/output,Control system,Distributed computing
Journal
Volume
ISSN
Citations 
8
2169-3536
0
PageRank 
References 
Authors
0.34
0
3
Name
Order
Citations
PageRank
Xiaogang Li100.68
Yiming Fang2309.71
Le Liu320.69