Neural Network Model for DC Spot Welding of Aluminum-alloy

The weldability of alumimum-alloy 508 for DC spot welding is studied by use of DC spot welder and dynamic parameter testing system. On the basis of technical testing, a discrete ANN model for quality prediction and evaluation of spot welding is built. The results show that the reflection from input vector space to output vector space could be realized by discrete processing the input and putput parameter of prediction model, such as welding current,voltage between electrodes and shearing strength. The prediction model has good ability of reliability and fault-tolerance. It is adapted to predict and evaluate weld quality for DC spot welding of aluminum,and can be used to realize the intelligent manufacture of resistance spot welding.