Online NIR diagnostic of laser welding processes and its potential for quality assuring sensor systems

摘要

We have integrated an imaging thermographic sensor into commercial welding optics for observation of the weld zone. Key element of the sensor is an InGaAs-camera that detects the thermal radiation of the welding process in the wavelength range of 1,200 to 1,700 nm. This is well suited to record images of the keyhole, the melt pool and the thermal trace. The sensor was integrated to the welding heads for on-axis observation to minimize the interfering contour to ensure easy adaption to industrial processes. The welding heads used were established industrial-grade TRUMPF optics: a BEO fixed optics with 280 mm focal length, or a TRUMPF PFO-3D scanner optics with 450 mm focal length. We used a TRUMPF TruDisk 16002 16kW-thin disk laser that transmits its power through a 200 μm core diameter light cable. The images were recorded and features of the various process zones were evaluated by image processing. It turns out that almost all weld faults can be clearly detected in the NIR images. Quantitative features like the dimension of the melt pool and the thermal trace can be derived from the captured images. They are correlated to process input parameters as well as to process results. In contrast to observation in the visible spectrum the NIR camera records the melt pool without auxiliary illumination. As an unrivaled attribute of the NIR sensor it supports an online heat flow thermography of the seam and allows identifying missing fusion ("false friends") of lap joints virtually during the welding process. Automated weld fault detection and documentation is possible by online image processing which sets the basis for comprehensive data documentation for quality assurance and traceability.