Deep learning-based quality prediction for multi-stage sequential hot rolling processes in heavy rail manufacturing

Xi Sun, Alessandro Beghi, Gian Antonio Susto, Zhimin Lv

Abstract: Highlights•We spatially align the isochronous data obtained from multi-stage rolling process.•A TFL model was crafted and effectively predicted the multiple quality indicators.•TFL use a Double-layer architecture to capture the transfer state between stages.•We introduce a Local Process Correlation Matrix to capture parameter correlations.•We use Multi-Gate Shared-Bottom architecture to enhance multi-objective prediction.