Development of a quality estimation model using multivariate analysis during resistance spot welding

Several researchers have studied the quality assurance of resistance spot welding using various methods. Because of the characteristics of the spot welding process, many problems arise when the weld quality is examined non-destructively in the in-process system. In order to solve such problems in this study, the process variables were monitored from the welding machine timer, from which it is possible to estimate weld quality in a real-time in-process system. For quality estimation, the factors were extracted as indices from the primary circuit dynamic resistance patterns, which were measured in the welding machine timer. The relationship between the indices and strength of the welds was observed through the correlation analysis. Using the indices, regression models that could estimate weld strength were obtained by regression analysis and factor analysis. Among the models, the most effective model was a stepwise-type regression model. Also, it is shown that the results of the proposed estimation models were consistent with the weld strength obtained through the experimentation.