Ruprecht-Karls-Universitšt Heidelberg

Automatic Process Control for Laser Welding


Mark Jaeger, Fred A. Hamprecht

Aim

In recent years industries increasingly use laser welding units. The benifit of laser welding compared to other methods lies in the higher economic efficiency.

Industrial laser welding is a highly dynamical and chaotic process and thus vunerable for process errors. Although process errors occure rarely, it is important to on-line monitore the process, to ensure that all faulty welds are detected. The aim of this study is to investigate classification strategies which allow a robust, automatic quality control of laser welding processes.

Methods

When laser radiation interacts with the work piece, secondary radiation is generated. This radiation contains information about the process stability and can thus be used to detect process errors.

The secondary radiation is recorded with one or several high speed CMOS cameras (with up to 40000 fps). Thus a high temporal correlation between the recorded frames is achieved. In state of the art laser welding process control systems the temporal dependencies are not fully exploited. Therefore we propose the use of Hidden Markov Models (HMM) to be able to model the temporal dependencies and increase the classification performance.

Experimental setup to monitor a welding process. The secondary radiation is recorded with a high speed camera.

State-space model of the laser welding process.

 



Last update: 06.10.2010, 12:27
zum Seitenanfang