Morgan Nilsen, doktorand inom produktionsteknik, håller sitt licentiatseminarium. Morgan är anställd och antagen till forskarstudier vid institutionen för Ingenjörsvetenskap. Titeln på licentiatuppsatsen är ”Optical detection of joint position in zero gap laser welding”.
Adjungerande professor John Powell, Luleå Tekniska Högskola
Professor Sven Goecke, Högskolan Väst
"This thesis presents an experimental study on how to find a narrow gap between metal sheets to be joined by laser beam welding. The use of laser systems for material processing is continuously increasing, and automated laser beam welding is gaining interest due to its ability to produce narrow and deep welds giving limited heat input and therefore less distortions compared to other processes, such as arc-welding. The automated laser beam welding process is however sensitive to how the high power laser is positioned with regards to the joint position. Deviations from the joint position may occur due to inaccuries of the welding robot and fixturing, changes in joint geometry, process induced distortions, etc. Welding with an offset from the joint position can result in lack of sidewall fusion, a serious defect that is hard to detect. This work evaluates monitoring systems for three different optical sensors; photodiodes, a camera and a spectrometer. A monitoring system is developed for three different photodiodes, one for the visual spectrum, one for the infrared spectrum, and one for the reflected high power laser light. The correlation between the signals from the photodiodes and the welding position is analysed using a change detection algorithm, and a correlation between the reflected high power laser light and the welding position is found. A camera with matching illumination and optical filters is integrated into the laser welding tool in order to obtain images of the area in front of the melt pool. This gives a relatively clear view of the joint position even during high spectral disturbances emitted from the process. By applying an image processing algorithm and a model based filtering method the joint position can be estimated with an accuracy of approximately 0.1 mm during normal welding conditions. By monitoring the spectral emissions from the laser induced plasma plume using a high speed and high resolution spectrometer, the plasma electron temperature can be estimated from the intensities of two selected spectral lines. The plasma electron temperature is correlated to the welding position and a relationship is found that can be used for finding the joint position."