• 목록

This study presents the development of a surrogate model using Artificial Neural Networks (ANNs) to predict the outcomes of a single pass in the H-beam Re-verse Tandem Mill (RTM) process, with the aim of replacing traditional Finite Element Analysis (FEA). By incorporating geometric conformity into the model, the ANN was able to accurately predict both displacement and material property changes across the cross-section of the H-beam, significantly reducing simulation time from tens of minutes to just a few seconds per pass. Two different ANN architectures were explored: a Fully Connected Neural Network (FCNN) and a ResNet model, with the latter showing superior performance, particularly in the Edger process. The results highlight the potential of ANN-based surrogate models for real-time optimization in the steel industry, offering a practical alternative to computationally expensive FEA simulations.