Quality control of resistance spot welding of high strength steels: Presentation held at Workshop MUSP - Sensors and data fusion in mechanical processes, April 10, 2013, Piacenza

A reliable multi-sensor measuring technique was developed for inline quality control of resistance spot welding consisting of a combination of sensory components in conjunction with multivariate analysis methods for monitoring the welding process. To achieve the best possible discrimination of the classes, a Fast Independent Component Analysis Algorithm (Fast-ICA) was applied. For the actual classification of the process data a multi-layer perceptron was used. The evaluation of the process and its quality was done on the basis of the process variables like welding current, voltage, time, resistance and forces. By the simultaneous evaluation of resistance and force profile during welding the probability of failure detection was increased to over 95 %.