| Welding is an important and key process in the manufacturing of marine engineering equipment that determines its quality.In our country major breakthroughs in technological innovation has been made in our country’s mainstream ship types,high-tech ships,marine engineering equipment and other fields,but there is still a big gap compared with the world’s marine equipment manufacturing powers.Intelligent manufacturing is the inevitable trend of marine equipment welding development.The bottleneck problem is to realize some key technologies of intelligentized robotic welding(IRW)of midthick plate,such as weld type identification,autonomous guidance,real-time weld seam tracking and on-line planning and real-time correction of multilayer and multi-pass(MLMP)welding.The current status of the welding process in our country’s ship-building is mainly reflected in the low degree of automation and low productivity.A large number of mid-thick plate welding is used in the ship-building section manufacturing,its filling volume is large,and the thermal efficiency of artificial gas metal arc welding(GMAW)is relatively low.The reason why the degree of automation is difficult to improve is that,on the one hand,it is difficult to guarantee the accuracy of the preprocessing and assembly error,which makes it difficult for robotic welding to be carried out in batches and to ensure its quality stability;on the other hand,in the process of MLMP welding of mid-thick plates,the existing offline planning software cannot offer quantitative guidance for actual welding,and repeated teaching is required even if robotic welding is used.Welding in ship segment manufacturing has a typical demand for IRW technology,which includes autonomous identification of weld types,self-directed guidance of welds,real-time weld seam tracking,MLMP on-line planning and real-time correction,which is the ideal application background to verifies the IRW technology.Based on the demands of the welding process in the ship-building manufacturing,aiming at the common technical problems for IRW of midthick plate robotic GMAW,some key technologies of IRW are studied in this paper,including autonomous welding guidance,real-time weld seam tracking,and adaptive welding path planning.In the mid-thick plate MLMP welding,the teaching-free weaving weld seam tracking is realized using the laser vison sensing technology.An on-line planning and real-time correction model of MLMP welding is established tosolve some of the pain points and difficulties in the application of robotic welding of mid-thick plates in ship building industry,and provide scientific verification and technical realization methods for the IRW in the multi-segment manufacturing process of ships,which is useful for improving the level of ship equipment manufacturing,and is of great significance to improve the independent supporting capabilities of shipbuilding engineering.First,a laser vision sensor is designed and developed,which can realize functions of the welding seam type recognition,weaving welding seam tracking,welding bead scanning and MLMP planning.And then,the quantitative relationship between the optical center distance,angle of laser,and height accuracy is analyzed,whilst,a fast-united calibration method based on synchronous image acquisition that can calibrate the internal and external parameters of the sensor and robotic system is proposed,which greatly improves the calibration efficiency of the vision sensor and robotic system,by70% compared to that of traditional methods.Subsequently,a typical welding classification method based on a priori model is proposed for the laser vision images of typical welding grooves.Aiming at the welding noise,such as spatter and specular reflection,that appears in laser vision images during the welding process,a noise removal algorithm based on the angle filtering and horizontal coordinate filtering of Steger method is proposed.Meanwhile,a dynamic line-shaped region of interest(DLROI)model is proposed to deal with the problem of moving features in the image during weaving welding process.The laser stripe on the previous frame is used as a reference for the segmented line-shape,and an area that has the highest structural similarity tothe reference line-shaped area is searched in current image.Moreover,through trajectory fitting,the actual phase of the weaving cycle can be obtained,and combined with the trajectory points stored in an intermediary array that have been extracted but not executed,the coordinate error during weaving welding process can be eliminated,so as to achieve real-time teaching-free seam tracking of the weaving welding.Next,the influence of welding parameters on weld bead parameters of the MLMP welding is analyzed,and the semi-quantitative mapping relationship of welding parameters(wire feeding rate,weaving amplitude,and side pause time)to welding quality and welding bead parameters is given.The weld groove is scanned using the laser vision sensor,and the point cloud data of the weld groove and weld bead are extracted.After rotating,slicing,projecting and denoising,the two-dimensional point cloud data of the groove section are extracted from the three-dimensional point cloud data.Furthermore,weld bead parameters are defined after quadratic curve fitting on the weld contour.The geometric profile feature data of the initial weld groove are taken as the input,and the welding process parameters of each layer and the Tool Center Point(TCP)pose and position are taken as the output.Then,through the method of model-scan-plan-weld-scan-correction,the robot GMAW MLMP on-line planning and real-time correction strategy is proposed based on the ideal welding groove model,filling bead type classification,welding parameters,and welding bead profile feature parameters.According to the strategy,an online planning and real-time correction model for MLMP welding for mid-thick plates is established.A comparison with low-magnification macroscopic measurement results of welded joints indicates that the height and width errors of the weld bead obtained by scanning planning are controlled within a range of ±0.6mm.Finally,according to the MLMP planning model,welding parameters and TCP pose and position for each pass of the V-groove welding test plate with a thickness of 20 mm are planned.Experiment result show that the seam tracking accuracy of the root pass weaving weld is within ±0.17 mm,and welding parameters and TCP position can be corrected on line.No welding defects such as cracks,incomplete fusion,or porosity occur across the seam.The applicability of on-line planning and real-time correction methods for MLMP welding based on laser vision is proven.It provides a scientific verification and technical realization path for the IRW of mid-thick plate in the subsequent ship multi-segment manufacturing process. |
PDF Full Text Request