Inteligência Artificial Aplicada a Processos de Dobragem de Tubo

Rotary Draw Bending is one of the most commonly used industrial processes for bending metal tubes, with applications in several industries. However, it presents significant challenges related to defects such as wrinkles and ovalization, as well as the natural elastic recovery of the material after forming. To minimize such defects, the process parameters should be tuned to achieve a controlled process. In this context, the scarcity of experienced operators compounded by world market dynamics places considerable pressure on OEMs and end-users. This project intends to ease the use of these machines by fully digitalizing the process and using state-of-the-art machine learning models to make the machines "smarter" and easier to use. The proposed research project aims to address different challenges related to tube bending processes. These challenges involve predicting and compensating the material elastic recovery (springback), minimizing defects, and optimizing process parameters to improve efficiency and product quality. By leveraging advanced computational techniques, machine learning algorithms, and real-time data integration, the project seeks to provide innovative solutions to these challenges, improving the competitiveness and effectiveness of tube bending operations. The project introduces novel concepts and approaches by integrating real-time data with virtual data from numerical simulations, enabling more accurate predictions and control of the bending process. Additionally, it aims to develop advanced AI applications, including predictive models and optimization algorithms, to assist operators in defining process parameters to achieve a "first-time-right" solution and ensure defect-free production. Furthermore, interdisciplinary collaboration is emphasized, combining expertise in mechanical engineering, materials science, computational modelling, and AI technologies to address complex challenges in tube bending processes. The AiBend principal objectives are the following: - Integrate real-time machine data with simulation data - Develop a monitoring solution - Develop an optimization method for process parameter optimization - Evaluate our developments in a real industrial setting