IntroductionThe on-line monitoring of quality in welding processes isnowadays a key issue in different industrial scenarios. Real-timemonitoring systems based on different technologies have been pro-posed during the last years to enable, not only a reliable and fastquality assurance, but also a better understanding of these pro-cesses. Among these proposals we can find solutions based onthe monitoring of the electrical signals of the process, for exam-ple via feature extraction techniques (Zhang et al., 2014), machinevision systems that may require laser illumination and filtering(Abdullah et al., 2008) or allow the estimation of the keyhole andweld pool sizes and their correlation with the main process param-eters (Zhang et al., 2015). Infrared thermography has also beenused for the estimation of the weld bead width and depth of pen-etration in conjunction with hybrid soft computing approaches(Chandrasekhar et al., 2015) and even enabling automatic defect∗