Research On The Application Of Collaborative Robots In Seamless Rail Welding

The seamless line is the best choice to meet the requirements of high speed and heavy load of modern railways,and welding is one of the key technologies to eliminate rail joints.Because the welding process of the track laying on the wire still has raised bumps in the annular band after processing,the domestic rail welding manufacturers currently use hand-held grinders,artificial profile grinding machines and rail bottom grinding machines to process the part.The polishing process has the problems of long processing time and high labor intensity.Collaborative robots are intelligent equipment that is easy to integrate.They have the advantages of high accuracy,high efficiency,and high safety.Using collaborative robots instead of manual grinding can greatly improve the machining accuracy and work efficiency of workpieces.To this end,this thesis studies a method for grinding and polishing seamless rail welds based on collaborative robots.The main work of the thesis is as follows:(1)In order to use the collaborative robot to realize the online recognition of rail welds,pre-process the rail weld pictures and establish an offline image library.Firstly,the characteristics of the seamless rail weld image were analyzed to determine the basic geometry of the weld.Then,wavelet filtering and welding region extraction are performed on the weld image.Then,the number of samples is expanded by geometric transformation and the size of the weld image is normalized.(2)Aiming at the rapid identification and location of rail welds,a seamless rail weld image recognition algorithm based on deep learning is studied.First,the principle of the convolutional neural network and the Alex Net model are cited.Then,the software development environment and system hardware configuration based on Tensor Flow deep learning framework are introduced.Based on this,a weld image recognition system based on deep learning is researched and designed.(3)Aiming at the problem of uneven grinding and polishing depth caused by sudden changes in burr shape and size during the grinding and polishing of the welding seam by the collaborative robot,a hybrid robot welding seam grinding and polishing force position control system based on a generalized prediction algorithm was designed.First,the generalized prediction theory and principles based on parameter models are cited,and a generalized prediction algorithm based on grinding and polishing environmental parameters is designed,and the environmental model is identified and feedback corrected online.Then,the theoretical model of traditional force-position hybrid control algorithm is cited and its advantages and disadvantages are analyzed.Then,combining the generalized prediction algorithm and the force-position hybrid control algorithm,a force-position hybrid system for the collaborative grinding and polishing of rail welds is designed,and the proposed algorithm is simulated and verified.(4)Finally,MATLAB and ADAMS software are used to jointly simulate the seamless rail welding seam grinding and polishing method based on the collaborative robot proposed in this thesis.Experiments show that the method proposed in this thesis can meet the requirements of high stability and high precision grinding and polishing of rail welds.