Research On Weld Seam Recognition And Path Tracking Of Welding Robot System For Rubber Bridge Bearings

Rubber bridge bearings,widely used in the construction of transportation infrastructure,are an important component to connect the superstructure and substructure of bridges.As the transportation infrastructure construction has been intensified,there is a huge demand for rubber bridge bearings in Chinese market.However,there is no mature intelligent automatic production equipment,especially in welding work.Due to the special characteristics of the workpiece structure,manual assisted welding and demonstration welding are main approaches applied,and no real automatic welding is realized.Therefore,in the research of this paper,the vision-based rubber bridge bearing weld seam recognition and welding path tracking method is proposed,and the main points of this are as follows.For the problem of noise in the weld seam image captured by CCD,the noise was reduced by pre-processing using wavelet transform algorithm based on improved threshold function.By comparing different threshold segmentation algorithms,the Otsu threshold segmentation algorithm was selected to extract the main contour information of the weld seam.For the shortage of threshold selection for traditional Canny algorithm in edge detection,an adaptive Canny improved algorithm based on K-Means++ was proposed to improve the adaptivity.The experimental comparison analysis verified that the algorithm in this paper was not only effective but also better than the traditional one.The elliptic equation was used to fit the weld seam image and the mean method to extract the centerline to obtain the feature information of the weld seam,which provided data for robot trajectory planning and path tracking.The welding trajectory planning was carried out in Cartesian coordinate system.A PD-type self-tuning iterative learning control algorithm based on improved forgetting factor was proposed for the influence of errors caused by external disturbances and uncertainty factors of the welding robot arm in path tracking.The introduction of forgetting factor and gain self-tuning term in the algorithm not only sped up the convergence of the system,but also enhanced the robustness and control accuracy of the system.The convergence of the algorithm was analyzed based on the paradigm theory,and the effectiveness and superiority of the algorithm stated in this paper was verified by performing an experimental comparison for the traditional PID control algorithm and PD-type iterative learning control algorithm.A vision-based welding control system was designed and experimental studies were carried out.Model selection and design were conducted for mechanical structures such as welding robots,welding positioners and industrial cameras in the system.Using the program framework based on Open CV and MFC,the software of the welding control system was designed.Featuring the functions of humancomputer interaction interface,algorithm operation and data logging,the software realized the coordinated control between the robot module and the vision module.The experimental results show that the welding system was highly reliable and met the requirements of rubber bridge bearing welding.