Surface features are crucial for assessing welding quality because they serve as an intuitive depiction of the quality of the joint and have a major influence on welding strength. According to the characteristics of the refill friction stir spot welding (RFSSW) process and an analysis of the surface-state and internal morphology of RFSSW joints, a method of predicting the mechanical properties of RFSSW joints based on surface-state characteristics was proposed. In this paper, a laser-ranging sensor was used to characterize the surface state of RFSSW joints, and parametric characterization methods of the surface-state features of RFSSW joints were proposed. On this basis, a support vector machine was used to predict and analyze the fracture mode of RFSSW joints. The accuracy of the analysis of the test samples reached 95.8%. This paper provides a more efficient and convenient new method for the quality evaluation of RFSSW joints.
Keywords: quality evaluation; refilled friction stir spot welding; support vector machine; surface testing.