[17a-K506-1~11] 3.6 Laser processing

10:30 AM - 10:45 AM JST(1:30 AM - 1:45 AM UTC)

[17a-K506-6] A deep learning-based simulator for laser welding

〇NOBUO HARA¹, Wataru Arakizeki¹, Akihisa Nakahashi¹, Akira Kumagawa¹, Izuru Nakai¹, Yoshiro Kitamura¹, Toshihiko Wada¹, Toshio Otsu², Tsuabasa Endo², Hiroharu Tamaru³, Yohei Kobayashi² (1.Panasonic Holdings, 2.ISSP Univ. Tokyo, 3.IPST Univ. Tokyo)

Keywords:

laser welding,deep learning,simulator

We developed the simulator that can simulate manufacturing variations focused on laser welding. Using our proprietary deep learning network, the Meister Data Learner (MeDaL), it has become possible to evaluate the variations in process behavior and results caused by vatiations in the workpiece and the environment in relation to manufatcturing conditions. This network is also expected to be applicable to other manufacturing processes.