Rolling force prediction during FGC process of tandem cold rolling based on IQGA-WNN ensemble learning

Aiming to further improving the calculation accuracy of rolling force in the FGC process of tandem cold rolling, the exit thickness accuracy and exit flatness accuracy of the strip in the first few coils after the gauge changing rolling process, a rolling force prediction method based on IQGA-WNN ensemble learning is proposed in this paper in light of the large change of strip parameters and the unstable quality of coils during the FGC process. Firstly, the traditional QGA is improved by quantum variation to avoid falling into local optimal solution. Secondly, the improved QGA is used to optimize the initial parameters of the network to improve the prediction ability of WNN. Finally, the WNNs improved by IQGA are used as the base learners for ensemble learning and are effectively integrated through bagging algorithm to further improve the prediction ability of the model. The rolling force prediction model proposed in this paper is tested on a 1450-mm five-stand tandem cold rolling production line. The results show that, compared with the traditional rolling force model, the separate WNN and IQGA-WNN models, for the first three coils of strip after FGC rolling, the ensemble learning model obtains the minimum rolling force calculation error and strip exit thickness and flatness deviation, the accuracy and stability of the rolling process are improved significantly, which proves the feasibility and effectiveness of the model proposed in this paper.