Selection of limited and constrained compensation tables for five-axis machine tools

Machine tool geometric inaccuracies are frequently corrected through the use of compensation tables available in machine tool controllers. Each compensation table contains a set of values that determine the incremental change in the commanded position of an axis given the current positions of the axes. While a five-axis machine tool, for example, can have at most 25 compensation tables, most machine tool controllers limit the number of compensation tables that can be implemented and provide constraints on the combinations of compensation tables that can be utilized. This work presents an artificial intelligence-based methodology to select and populate the optimal set of machine tool compensation tables when these limitations and constraints exist. Using data from an industrial five-axis machine tool to construct a kinematic error model, simulation results for the proposed methodology and a heuristic based on the impact of individual compensation tables when selecting six compensation tables are compared, and the proposed methodology is found to outperform the heuristic. The proposed methodology and a solution based on a full set of compensation tables are experimentally implemented on the machine tool and the mean volumetric error resulting from the proposed methodology is found to be only 25 mu m less than the volumetric error resulting from the full set of tables. The proposed methodology is then implemented in two more simulation studies where constraints are imposed on which combination of compensation tables could be used and which type of compensation tables could not be utilized. The resulting mean volumetric error was 7.0 and 28.3 mu m greater, respectively, than the unconstrained solution.