Abstract:
With the advancement of computational resources and the increase in data generation and storage capacity, approaches based on artificial intelligence and machine learning (ML) have been continuously incorporated into industrial processes. This data-driven methodology has made it possible to map, in a reasonable way, the complex relationship that exists between industrial processing, the chemical composition and mechanical properties of steels. In this way, a tool based on ML was developed to predict the mechanical properties in tension, industrial processing conditions and chemical composition of advanced high-strength steels (AHSS), for automotive application, processed via Continuous Annealing Process Line (CAPL). An industrial database was used to generate information and knowledge about the materials evaluated. The accuracy of the prediction models was evaluated through error parameters, which are linked to the rationality and reliability of the tested method. The results showed that the XGBoost algorithm can be used to build different models, with excellent performance metrics. For the practical application of the tool, a program was developed for use in a web site, with a friendly interface, to assist in the development of new products and decisions related to the optimization of the processing route.