Deep Learning makes laser welding precise and safe

Among other things, the company Manz supplies modern laser systems for the production of lithium-ion batteries. To ensure precise and safe laser welding processes, the company uses image processing software with integrated deep learning technology. For example, a battery module with 90 cells, as used in electric vehicles, contains 540 welding spots, each 1 mm in size. The challenge is to create a safe and stable welded joint without damaging the comparatively thin and sensitive battery housing. Manz therefore relies on an inline inspection system for 100% control of the welding process. Among others the standard software Halcon of MVTec software is used. With the Deep Learning functions integrated therein on basis artificial intelligence (AI) the error recognition can be improved clearly. The technology lowers the programming expenditure significantly and gets along also with before not exactly defined errors. Especially the so-called anomaly detection requires only a small amount of image data for training. As a result, companies like Manz can save a lot of time and corresponding costs.

www.mvtec.com