A Study on Prediction of Welding Quality Using Mahalanobis Distance Method by Optimizing Welding Current for A Vertical-position Welding

Abstract The one of challenging position in the welding process is a vertical-position due to the difficulty. Compared to a horizontal-position, arc welding on the vertical-position is much more difficult because the metal transfer is influenced by the gravity force. Moreover, the gravity force leads to the welding quality decreased that caused by welding fault. To detect the welding fault, the method has still been used based on off-line method which has many disadvantages. One of the disadvantages is the welding fault detection can be performed after the welding process finished. Therefore to deal with this problem, this study is proposed the new algorithm based on Mahalanobis Distance (MD) method for on-line monitoring system. The experimental was carried out with 3 different setting of welding current in order to find out the optimal setting. From the experimental result, it proved that developed algorithm could achieved the highest welding quality at 250A welding current setting which the welding quality is 98.98% for the start section and 98.96% at the middle section. By additional experiment, It was verified that the developed algorithm based on optimized welding current could determine the good quality of weld.