Research Of Registration Parameter Calibration And Data Fusion Technology For Large Scale Measurement Field

The building of high-precision large scale measurement field is the foundation for the automatic assembly process of large parts.When measuring large parts,registration is always necessary for equipment to reach all of the key points.Among the existing calibration methods for registration parameters,the most common one is to match the coordinates of ERS points measured at different stations using Singular Value Decomposition algorithm.The rotation parameters and translation parameters can be solved directly but the accuracy of registration is relatively poor.The layout of the ERS points is decided depending on personal experience and optimization for the coordinates of ERS points can not be carried out because of lack of quantitative evaluation index.Besides,single kind of measuring equipment is used to complete the large size mesurement,and therefore measurement accuracy and range scale are difficult to take into account at the same time.Laser tracker and total station that are commonly used to build large scale measurement field are selected as the main research object.Firstly,the registration model is improved and the Jacobian matrix iteration method is adopted to solve the registration parameters.Secondly,quantitative evaluation index for ERS points layout was proposed and a hybrid intelligent algorithm is used to optimize the ERS coordinates.Next,high precision angle measuring unit of the total station is taken advantage of to realise data fusion to improve the measurement accuracy.At the same time,a combined measurement system including laser tracker and total station is established and high-precision large scale measurement is realized by data fusion between the two devices.The research work can be summarized as follows:1)The impact of ERS points deformation on regisration parameter calibration accuracy is analyzed and registration model is improved by introducing 3D deformation coefficient.Jacobian matrix iteration algorithm is used to solve the new registration model and the relationship between measurement error and registration parameter error was discussed.The condition number of Jacobian matrix was put forward as the quantitative evaluation index for ERS points layout.Simulation is conducted under different distance,quantity,shape and dimensions of the ERS layout,which indicates that the registration parameter error becomes larger along with increasing the condition number of Jacobian matrix.Furthermore,recommendations for the ERS points layout design are provided.Genetic simulated annealing algorithm is adopted to solve the local optimal solution problem during the optimization process of ERS coordinates.2)Based on the principle of laser tracker measurement network,redundant measuring data from total station in multiple stations is obtained and the result is optimized within the constraint of high precision angle measurment value.The measurement precision of the total station is significantly increased.A combined measurement system including laser tracker and total station is established.Matrix weighted linear minimum variance optimal fusion criteria is used to improve the measurement precision of the ERS points,and the combined measurement error is controlled within 1mm in simulation.3)Registration experiment is conducted with laser tracker and the result indicates that the improved model and solving method in this paper can transfer the coordinate of test points between different stations more accurately compared with the traditional algorithm.The registration accuracy becomes lower with the increase of the condition number of Jacobian matrix related to the ERS layout.The method is applied to the pose measurement of the moving platform of the three DOF translational parallel mechanism with the accuracy on the scale of 0.1mm.Finally,total station data fusion experiment is carried out and the measurement precision of the end surface of the large part is increased by nearly 30%.In conclusion,this thesis provides an improvement plan for the registration parameter calibration process in large scale measurement and demonstrates a novel method for improving the measurement accuracy using data fusion algorithm.Additionally,experimental study is also carried out to verify the rationality of the calibration method based on Jacobian matrix iteration and the data fusion algorithm for total station.It has been proven that our method improves the overall precision of large scale measurement significantly,attracting certain practical applications and leading to great assistance to the automatic assembly process of large parts.