Quantitative grinding depth model for robotic weld seam grinding systems

Abstract

The service cycle and dynamic performance of structure parts are affected by the weld grinding accuracy and surface consistency. A flexible abrasive belt wheel used to grind a weld can avoid burns on the base material and improve the grinding efficiency. However, owing to the low positioning accuracy of the robot, elastic deformation of the abrasive belt wheel, and uneven height of the weld, the actual weld grinding depth cannot be determined, which results in scraping or reduction in the service cycle of the structural parts. Therefore, the robot weld grinding system requires an accurate quantitative model of weld grinding depth. Considering these problems, in this study, a robot weld quantitative grinding depth model is proposed, considering the effect of the actual contact process of abrasive particles and robot positioning error in the grinding depth direction. The contact contour is determined as a rectangle based on the simulation contact model, which is different from the common ellipse contact area. The weld removal depth model based on contact pressure and a single abrasive grain trajectory is then established. Subsequently, considering the positioning error of the robot grinding depth direction, the kinematic parameters of the weld grinding robot are calibrated based on differential transformation and iterative process of a linear equation to obtain the compensation amount of the robot grinding depth direction. Comparative experiments further verified the accuracy and effectiveness of the system. The average weld height after grinding by the model was 0.15 mm, while the maximum was 0.18 mm, which reduced the error of weld removal depth by approximately 50 % compared to the traditional approach. The roughness Ra could reach 0.502 μm.