“…In laser-based manufacturing field, such data-driven approaches have been extensively studied in the past and are based on autoregressive exogenous (ARX) model [154]- [155], cluster analysis [156], fuzzy logic (FL) [156]- [161] or on supervised learning algorithms including multivariate regression (MR) [162]- [163], multi-layer perceptron (MLP) [164]- [165], and decision trees (DT) [166]- [167], as well as K-nearest neighbors (KNN) [168]- [169]. Once the eigenvector has been established, effective identification and classification of different welding status or defects can be realized by using advanced modeling technology.…”