Identification of multiclass defects in aluminum alloy resistance spot welding based on support vector machine

A model is built to identify splash defect and incomplete fusion defect of aluminum alloy resistance spot welding based on Support Vector Machine method.The characteristic vector used in the model is extracted from process curves of aluminum alloy resistance spot welding.This model is trained and tested with different sample data.The test result shows that the accuracy rate of identifying splash defect is 96.7% and the accuracy rate of identifying incomplete fusion defect is 100% under given sample data.Therefore, it is reliable to identify multiclass defects of aluminum alloy resistance spot welding with Support Vector Machine method.