The cross-sectional profile quality of medium- and high-carbon steels is crucial in high-end manufacturing. Machine learning algorithms are employed to predict transverse thickness differences in cold rolling. Facing variable-length sequences in single-stand mills, the predictive abilities of multi-layer perceptron, random forest, long short-term memory and transformer models are compared, with the transformer model showing the highest accuracy, with an R2 of 0.9783. Then, machine learning is integrated with particle swarm optimisation, a cold rolling mechanism model and stable rolling requirements to create a rolling schedule optimisation model aiming to minimise transverse thickness differences. A finite-element model of the S6-High cold rolling mill (CRM) is established via ABAQUS to interpret optimisation results. Field applications reveal that this research reduces the average transverse thickness difference in rolling medium- and high-carbon steels from 26.6 to 17.5 μm, greatly improving the control level in the S6-High CRM.
LiYHHoustenWZhaoY,et al.Cu/single-walled carbon nanotube laminate composites fabricated by cold rolling and annealing. Nanotechnology2007; 18: Article 205607.
2.
BaiJLPanHZ. Study on flatness control strategy of a single-stand reversing cold rolling mill. Adv Mater Res2014; 1004: 1193–1197.
3.
SongCCaoJWangL, et al.Transverse thickness profile control of electrical steel in 6-high cold rolling mills based on the GA-PSO hybrid algorithm. Int J Adv Manuf Technol2022; 121: 295–308.
4.
WeiZWangGSunJ, et al.Development of cold-rolling edge-drop control technology for electrical steel: a review. Steel Res Int2023; 94: Article 2200919.
5.
LinghuKJiangZZhaoJ, et al.3D FEM analysis of strip shape during multi-pass rolling in a 6-high CVC cold rolling mill. Int J Adv Manuf Technol2014; 74: 1733–1745.
6.
WangXYangQHeH, et al.Effect of work roll shifting control on edge drop for 6-hi tandem cold mills based on finite element method model. Int J Adv Manuf Technol2020; 107: 2497–2511.
7.
HeHShaoJWangX, et al.Silicon steel strip profile control technology for six-high cold rolling mill with small work roll radius. Metals (Basel)2020; 10: 401.
8.
HanGLiHWangG, et al.Prediction and control of profile for silicon steel strip in the whole tandem cold rolling based on PSO-BP algorithm. J Manuf Process2024; 120: 250–259.
9.
YuanTSunWHeA, et al.Research on quarter wave shape control of cold rolled high strength steel based on BP-MOPSO. Int J Adv Manuf Technol2022; 122: 3867–3880.
10.
WangYLiCJinX, et al.Multi-objective optimization of rolling schedule for tandem cold strip rolling based on NSGA-II. J Manuf Process2020; 60: 257–267.
11.
LövbergA. Remaining useful life prediction of aircraft engines with variable length input sequences. In Annual conference of the PHM society (2021, December) (Vol. 13, No. 1). https://doi.org/10.36001/phmconf.2021.v13i1.3108
12.
BashiriFSCareyKAMartinJ, et al.Development and external validation of deep learning clinical prediction models using variable-length time series data. J Am Med Inform Assoc2024; 31: 1322–1330.
13.
OkaYSudohKNakamuraS. Length-constrained neural machine translation using length prediction and perturbation into length-aware positional encoding. J Nat Lang Process2021; 28: 778–801.
14.
JiYFSongLBSunJ, et al.Application of SVM and PCA-CS algorithms for prediction of strip crown in hot strip rolling. J Cent South Univ2021; 28: 2333–2344.
15.
WangDXuYZhangT, et al.Quantitative study on the relationship between the transverse thickness difference of cold-rolled silicon strip and incoming section profile based on the mechanism-intelligent model. Metall Res Technol2021; 118: 3.
16.
ChenSZZhangXPengLG, et al.Multi-objective optimization of rolling schedule based on cost function for tandem cold mill. J Cent South Univ2014; 21: 1733–1740.
17.
YoonSHYuHJ. A simple distortion-free method to handle variable length sequences for recurrent neural networks in text dependent speaker verification. Appl Sci2020; 10: Article 4092.
18.
ZhangZKrawczykBGarciaS, et al.Empowering one-vs-one decomposition with ensemble learning for multi-class imbalanced data. Knowl Based Syst2016; 106: 251–263.
19.
BittencoutFRZárateLE. Hybrid structure based on previous knowledge and GA to search the ideal neurons quantity for the hidden layer of MLP – application in the cold rolling process. Appl Soft Comput2011; 11: 2460–2471.
20.
YangWQZhaoZTZhuLY, et al.Strip flatness prediction of cold rolling based on ensemble methods. J Iron Steel Res Int2024; 31: 237–251.
21.
ChenYPengLWangY, et al.Prediction of tandem cold-rolled strip flatness based on attention-LSTM model. J Manuf Process2023; 91: 110–121.
22.
VaswaniAShazeerNParmarN, et al.Attention is all you need. Adv Neural Inf Process Syst2017; 30: 5998–6008.
23.
DingCYYeJCLeiJW, et al.An interpretable framework for high-precision flatness prediction in strip cold rolling. J Mater Process Technol2024; 329: 118452.
24.
JiYWenYPengW,et al.Predicting hot-rolled strip crown using a hybrid machine learning model. ISIJ Int2024; 64: 566–575.
25.
LiXFangYLiuL. Kernel extreme learning machine for flatness pattern recognition in cold rolling mill based on particle swarm optimization. J Braz Soc Mech Sci2020; 42: 70.
26.
ChenLSunWHeA, et al.Research on thickness defect control of strip head based on GA-BP rolling force preset model. Metals (Basel)2022; 12: 24.
27.
SunWChenLHeA, et al.Study on quarter-wave control of DP980 steel based on a bending force linear combination strategy. Int J Adv Manuf Technol2023; 127: 2881–2892.
