A fuzzy pattern recognition based system for monitoring laser weld quality

In-process monitoring of welding has become important as the use of laser w elding increases. Plasma and spatter are measured and used as signals for e stimating the quality of a weld. The measurement system consists of three p hotodiode sensors (one IR and two UV) to detect the plasma and spatter sign als in CO2 laser welding. The estimating algorithm was constructed using fu zzy pattern recognition considering the amplitudes as well as amounts of da ta beyond the tolerance boundary. Weld qualities were classified as optimal heat input., slightly low heat input, low heat input and misalignment of f ocus. Also, an algorithm for detecting spatter was created in order to find the partially produced pit. These algorithms were used for quality monitor ing in tailored blank welds with a CO2 laser.