Study On Multi-parameter Modeling Method Of Articulated Coordinate Measuring Machine

Traditional coordinate measuring machines gradually can not meet therequirements of fast and online measurements. The non-orthogonal coordinatemeasuring machines have been extensively developed and widely used because oftheir flexibility, easy operation and online measurement in recent years. Thedissertation studies on a kind of the non-orthogonal Articulated CoordinateMeasuring Machine (ACMM) around modeling method and model parametersidentification.The well-known Denavit-Hartenberg (DH) kinematic modeling method hasbeen often used for modeling of ACMM, but the DH matrix is morbid when it isapplied to represent parallel joints. Therefore, the other two modeling methods, afive-parameter method and an improved four-parameter model, have been studiedin response to this ill-conditioned matrix problem in the dissertation. After analysis,the improved four-parameter model combined with DH model is put forward for theACMM modeling under study. There are three probe-stations in the ACMM and ineach case the dissertation presents a kinematic model contained certain kinematicparameter number and different meanings of probe parameters. The dissertationalso makes a comprehensive error analysis for the ACMM under study and builds asystem error model. Another kinematic model contained joint-shaking errors havebeen established because of its significant impact on measurement accuracy. Basedon the multi-parameters kinematic model, Gauss-Newton method is applied forcalibrating the model parameters. Redundant parameters of the kinematic model areeliminated in the calibration process and the best estimates of the kinematicparameters are found after several Gauss-Newton iterations.By experimental verification, this kind of modeling method combinedimproved four-parameter model with DH model can describe the movement ofArticulated Coordinate Measuring Machine with parallel joints. The best estimatesof the kinematic parameters are brought back to kinematic model and severalrepeated measurements of standard bar show that the accuracy of the ACMM under study has been greatly improved.