Abstract: Automated real-time characterization of resistance spot welds using ultrasound-based nondestructive evaluation requires a computational process and system to accurately and rapidly interpret the ultrasonic data in real time. Such a process can be automatically learned using artificial intelligence, from a dataset of exemplary ultrasonic data from nondestructive evaluation of resistance spot welds for which a corresponding ideal evaluation of each weld is provided. The process can then be implemented into a system to automatically interpret data from non-destructive evaluation in real-time. The ideal evaluation of each weld requires identification a large set of features that are observable in the ultrasonic signature and comprehensively characterize the corresponding weld process.

Abstract: A system for resistance spot welding control and a method thereof are disclosed. A system for regulating welding parameters of a spot welding machine joining a plurality of panels according to an aspect of the present disclosure includes an ultrasonic sensor installed inside an electrode of a welding gun, applying a ultrasonic wave to a welding part of the panel, and detecting a reflected ultrasonic signal; an ultrasonic analyzer capable of creating ultrasonic analysis information by analyzing in real time the ultrasonic signal; and a welding controller receiving the ultrasonic analysis information from the ultrasonic analyzer, performing adaptive welding control with the welding parameters set based on information of the panel, and compensating in real time one or more welding parameters according to the received ultrasonic analysis information.