PR-DT1021-1 - 2021 Digital Transformations, October

Synthetic Images of Longitudinal Cracks in Stainless Steel Slabs via Wasserstein Generative Adversarial Networks Used Toward Unsupervised Classification

D. Andrade, M. Simiand, J. Barreiro

Download PDF

View Sample Page

Abstract:
Primary stainless steel production for an integrated steel mill includes ironmaking, steelmaking, casting, rough rolling and product rolling. While all of these steps have their inherent problems, this paper will focus on those manifested in the slabs after the cutting torch. Surface defects on continuously cast slabs require treatment by grinding. This new phase in the process causes lower throughput of the final product and additional energy costs. The defects during rolling are carried out through the entire line of production when not detected early. The problems do not arise immediately; the fractures or dislocation continue with the rolling downstream. Where smaller and smaller thicknesses make them appreciable, they can waste significant time and money for any given production line. A low-tech, low-cost, effective solution is a visual inspection of the slab surfaces during the production along with progressive steps during rolling, which requires constant conditional awareness by qualified operators.


Keywords: digital transformation, industry 4.0