A Statistical Approach to Nondestructive Testing of Laser Welds

Abstract

A statistical analysis of the data obtained from a relatively new nondestructive technique for laser welding is presented. The technique is one in which information relating to the quality of the welded joint is extracted from the high intensity plume which is generated from the materials that are welded. The system is such that the detected plume is processed to give a numerical value associated with the material vaporization and consequently, the weld quality. Optimum thresholds for the region in which a weld can be considered as acceptable are determined based on the Neyman-Pearson criterion and Bayes rule.