Wavelet analysis of light emission signals in laser beam welding

Abstract

This paper introduces a compact, robust, and versatile generation of a laser processing head coupled with a single-mode Ytterbium source StarFiber 150 P which not only is used in a wide range of laser material processing applications but also enables in situ and real-time quality control in production lines. In this study, the effect of power density on the weld quality of bead on plate in a SHADOW (Stepless high speed accurate and discrete one pulse welding) welding process was considered. Furthermore, the co-axial back reflected/emitted light from the process zone was measured using three optical sensors; each one measures the light emission in a certain waveband (450–850, 1000–1200, and 1250–1700 nm). From these sensors, the recorded signals were first used as benchmarks to correlate with the weld quality features and defects. Discrete wavelet transform has been used to analyze the signal within the time–frequency domain. The detail signals at different decomposition levels disclosed information with regard to the type (mode) and dynamics of welding process. Accordingly, the deep penetration welding (keyhole) was confidentially distinguished from the shallow (conduction) welding mode. Several defects such as process instabilities (high fluctuations in the keyhole dynamics), formation, and collapse of the keyhole, as well as the lack of penetration, were detected from the corresponding signal signatures. In addition, this information permits us to locate the defect/transition site through the in situ and real-time quality control of the process in production lines. In contrast, porosity and cracks could not be sensed.