A STUDY ON ERROR COMPENSATION FOR CNC MACHINE TOOL

Abstract

This thesis studies the error compensation approach for CNC machine tools. The objective of the project is to develop a feasible error compensation approach so as to efficiently predict and compensate the geometric/kinematic errors of CNC machine tools. In the thesis, the error sources of the machine tool in machining process are first analyzed and classified. A universal geometric/kinematic error model of machine tools is then derived according to the rigid body kinematics. An experimental procedure is proposed and a group of test pieces are carefully designed and machined to identify the derived error model for a CNC lathe. The machine errors are derived from measurements of the workpieces using a 3-axis coordinate measuring machine. A neural network approach has been developed to generate the compensated toolpaths based on the error model. A curve fitting method, i.e. biarc interpolation, has been adopted to fit the compensated tool-path for final machining. The generation of the corresponding CNC codes from the compensated tool-path has also been discussed after the curve fitting. All the algorithms in the proposed error compensation approach have been implemented on a personal computer.