| At present, offshore pipeline is widely used in oil and gas storage and transportation, and it is in great demand. However, due to the level of pipeline welding is still not developed in China, semi-automatic welding is still the most common method in offshore pipeline welding, which leads to long welding cycle, low efficiency and poor safety. Therefore, the key problem in pipeline welding is how to realize the full automation in the whole process. On the basic of analyzing the present situation of pipeline welding, and summarizing the research status of machine vision in welding field, this paper researches the key technologies of the pipeline welding robot based on visual sensor.Firstly, the offshore pipeline welding system based on visual sensor is established, especially the design of the visual system is described in detail, and then the calibration of visual system is executed. On this basis, in this paper, the positioning of initial welding position is studied, and a method of positioning the initial welding position automatically for offshore pipeline welding is proposed. In the proposed method, Pulse Coupled Neural Network is adopted for detecting the initial welding position in real time during the synergic movement of robot and positioners, and extracting the pixel coordinate of initial welding position. Then based on binocular vision, the three-dimensional coordinate of initial welding position is calculated, and the feasibility of this method is verified by experiments.Secondly, the three-dimensional reconstruction of pipeline weld seam is researched deeply. The sub-pixel coordinate of weld seam is extracted based on Zernike moment. On this basis, to realize the three dimensional reconstruction, the sub-pixel points of the same weld seam but in two images are matched uniquely based on epipolar constraint. Then experiments are carried out to verify the feasibility and robustness of the sub-pixel coordinates extraction and three-dimensional reconstruction. Besides, the design of control software for the robot welding system in this paper is introduced emphatically, and especially the design of production data management module is described in detail. Finally, the whole welding system is simulated in the Robot Studio, and actual welding is executed by using the above methods.Finally, the main work done in this paper is summarized, and then the further work and the research directions in the future are discussed. |
