AI System For Profile Production Simulation

Simulation And Measurement Of Deformation Field Of Large Deformation Section Steel Rolling Thermal + Mechanical Simulation

When the project leader first worked as a doctoral project in Germany, he completed a large-scale project with a total investment of 1 million German marks by the German national DFG (German Research Association). This project pioneered the factory application of finite element artificial intelligence model technology in the production process of large-scale complex section profiles, final shape prediction and production process optimization. I reduced the simulation time of a processing process from three days and three nights to 20 hours, and greatly improved the accuracy of prediction; In particular, a simplified model is developed, which can be calculated in only 5 ‰ (5 ‰) time, and the assumptions of the simplified model can be coordinated with today's machine learning.

DFG (Deutsche Forschungsgemeidschaft, i.e. German Research Association) of German national scientific research set up this project to verify whether the finite element simulation technology can be applied at the factory level and used to simulate the section steel rolling system of large deformation heat + mechanics after decades of industry research and development! Therefore, high temperature rolling of angle steel based on six pass design, continuous casting and rolling of H-beam, rail production and so on are selected. All rolling is carried out by hot rolling with real steel (unlike using plastic mud to simulate hot steel in China), and in the section steel rolling of angle steel, grids are milled on the surface and inside to measure the deformation field. In order to study the mesh, plate rolling was carried out first; Then a special pass design is adopted so that the deformation field can be measured, because the grid can only withstand up to three times of rolling! To this end, only the project experiment has trained a master and several undergraduates!

The simulation of the project is more difficult! Everything should use real data, and the mechanical properties (strength, elastic modulus and Poisson's ratio, etc.) and thermal properties (heat capacity, heat transfer coefficient, etc.) at high temperature all change with temperature and change exponentially within the deformation range! Moreover, rolling pulls the rolled piece into the complex pass through friction, which is much more complex than forging or flat roll rolling! ANSYS was incompetent at that time, so we had to use Marc! The simulated deformation field data are compared with the measured deformation field. In addition, the overall deformation and overall force energy are compared.

See Summary of Ph.D. thesis topics.

The project achievements were first applied in Germany, the largest equipment technology company in the industry, and were also invited to Danieli, the largest technology company in the industry in Italy for technical exchange on three occasions; Finally, the main developer received an employment contract from Morgan, the largest technology company in the industry in the United States, before graduation. In addition to several of Europe's largest technology companies, there are letters of invitation from several R & D institutions in Europe and the United States (including the joint invitation of three directors of a European University for three times!). This is the first time that we have accomplished something that our peers think is impossible (there are two more times since then). This research achievement has contributed to the publication of the team's first book and the online publication of the name of the main researcher for eight years!

At that time, the metal forming Research Institute established a four continuous rolling line bar mill with a rolling speed of 70 meters per second in the 1980s. After decades of research and development, there were more than 40 doctoral theses on line bar rolling. In the United States, Morgan completed three groups of more than 100 sets of models, part of the data comes from this. The pass design software developed based on these models has been further optimized in the project of POSCO in South Korea, and has been fully recognized by TISCO and other enterprises in China.