Real-time welding condition monitoring by roughness information extracted from surface images

Abstract

Monitoring the friction stir welding is an important part of intelligent systems. In this paper, we designed welding experiments based on the single factor variable method and studied the roughness of the weld surface in detail. By analysing the variation in the roughness, the theoretical formulas of roughness Ra and Rsm are obtained. The deviation between the measured value and the theoretical value of Ra can be used to evaluate the welding condition. An industrial camera was used to capture the texture images of the weld surface and process them to extract features. Based on the fractional Brownian motion model, the fractal dimension of statistical significance was obtained from the power spectrum of the gray distribution in texture images. The results show that the extracted fractal dimension can accurately reflect the changes in roughness Ra. This research is helpful to realize welding condition monitoring based on machine vision systems.