Abstract:

Several of the dynamic electrical parameters of alternating current resistance spot welding are blended by use of artificial neural networks,and a monitor system of spot welding quality of mild steel is established in this research.In this system,the dynamic electrical parameters are used as input space and the sizes of nugget are used as output space.The system can be used for detecting the quality and forecasting the size of nugget on real time during resistance spot welding. The average forecasting error of diameter of nugget is less than 5% and the average forecasting error of height of nugget is less than 8% in this monitor system. The system can satisfy the actual need of engineering completely.

Key words: artificial neural networks, resistance spot welding, quality control