A method, apparatus, equipment and medium for predicting the remaining service life of rolling mill rolls

Abstract

This invention discloses a method, apparatus, equipment, and medium for predicting the remaining service life of rolling mill rolls, relating to the field of rolling mill roll detection technology, and addressing the problem of low accuracy in online prediction of the remaining service life of rolling mill rolls in existing technologies. The method includes: acquiring an initial rolling mill roll life dataset, comprising at least an image of the rolling mill roll surface to be predicted and a vibration signal of the working mill to be predicted, during the online phase; employing a constructed multi-source information fusion model, ResNet18-DA, to extract and fuse features from the rolling mill roll life dataset to be predicted, obtaining target features; and determining the prediction result of the remaining service life of the rolling mill roll. This invention enhances prediction accuracy and robustness by fusing data from the rolling mill roll surface image and the working vibration signal, establishing a ResNet18-DA model, determining the relationship between the rolling mill roll wear image and the working vibration signal and the remaining service life of the rolling mill roll, and achieving accurate online prediction of the remaining service life of the rolling mill roll.

Images