Accuracy and calibration of conventional and parallel link machine tools.

摘要

Dimensional accuracy is the single most important criterion used to evaluate the quality of machined parts. Increasing demand for highly accurate and inexpensive mechanical parts has led to the development of economical and efficient processes for their production. One crucial factor affecting the quality of machined parts is the accuracy of the machine tool used to produce that part. Software error compensation is one effective method for improving the accuracy of machine tools. This requires the knowledge of the machine's errors derived from empirical models or theoretical models that have been designed to either update the kinematics of the machine via software or to update the numerical control (NC) code.; The steps involved in achieving a successful software calibration scheme for a machine tool are as follows: (1) Error Modeling, (2) Error Observation, (3) Model Parameter Estimation, and (4) Error Compensation (software). This research addresses each, with particular emphasis on solution techniques for the first three steps.; The error modeling addressed in this research is associated with parallel link machine tools. A kinematic error-model for a Stewart Platform-based machine tool is presented. The error analysis shows that forty-two kinematic parameters completely characterize the errors for a Stewart Platform machine tool. This error model allows for decoupling these forty-two parameters into six problems each with seven error parameters. Transmission error bounds are also developed for the Stewart Platform machine tool. These methods can be expanded to other types of parallel link machine tools.; The error observation portion of this research uses a novel gauge named the 'Plate-Bar Mechanism'. This gauge utilizes a ball-bar mechanism and is capable of measuring five error components in a circular path within the workspace. The system measures thermal errors and dynamic errors in three-axis machining centers. A simple trilateration along with a long stroke LVDT (linear variable differential transformer) is also used to measure three error components in the workspace of a parallel link machine tool.; Least squares estimation is the most common technique used for parameter identification of machine tool error models. This research presents a new criterion which identifies the parameters based on the maximum error in the workspace. Experimental results show a 25% reduction in the maximum error over the least squares criterion.; The error modeling, error observation, and error estimation techniques are then implemented on a three-axis parallel link machine tool. The final results show that successful calibration of a parallel link machine tool is possible when the research presented in this thesis is implemented.