Abstract— The welding process is a very common thing done in industrial activities. The welding results must certainly be assessed so that the quality of the welding results is guaranteed. The quality of the welding results can be seen from the absence of defects on the surface of welding area. This research aims to determine the quality of welding results. The methods used to identify welding defects in penetrant test results are the Grey Level Co-Occurance Matrix (GLCM) and K-Nearest Neighbor methods. From 80 test data, total accuracy reached 85% which is divided into 4 classifications namely good, porosity, transverse and longitudinal. Test results for the good class obtained the highest accuracy of 90% of the 20 good test images compared to the accuracy of the other 3 classes
Copyrights © 2022