New metrology tools, such as laser trackers, are enabling the rapid collection of machine tool geometric error over a wide range of the workspace. Error models fit to this data are used to compensate for high-order geometric errors that were previously challenging to obtain due to limited data sets. However, model fitting accuracy can suffer near the edges of the measurable space where obstacles and interference of the metrology equipment can make it difficult to collect dense data sets. In some instances, for example when obstacles are permanent fixtures, these locations are difficult to measure but critically important for machining, and thus models need to be accurate at these locations. In this paper, a method is proposed to evaluate the model accuracy for five-axis machine tools at measurement boundaries by characterizing the statistical consistency of the model fit over the workspace. Using a representative machine tool compensation method, the modeled Jacobian matrix is derived and used for characterization. By constructing and characterizing different polynomial order error models, it is observed that the function behavior at the boundary and in the unmeasured space is inconsistent with the function behavior in the interior space, and that the inconsistency increases as the polynomial order increases. Also, the further the model is extrapolated into unmeasured space, the more inconsistent the kinematic error model behaves.

This content is only available via PDF.
You do not currently have access to this content.