A digital twin model of hot rolling process based on CGAN

Aiming at the current shortcoming of relying on online control of process parameters in hot rolling process, we proposed a digital twin model based on CGAN. This model can simulate the rolling process according to rolling information such as production conditions, working environment, material properties, and rolling sequence. For model training, testing and validation, we collected actual production data of the hot rolling mill is collected. The results show that the process parameters generated by the model are close to those of the real process, which proves it can reflect the hot rolling process and could be used for future simulation, quality prediction and production optimization of the hot rolling process.