A Strategy for the Compensation of Errors in Five-Axis Machining

This paper outlines a strategy for error compensation in five-axis machining. The strategy uses a neural network model for compensation of errors due to changes in temperature and axis motion. Long term errors due to wear will be addressed with an adaptive approach. A method based on the analysis of the neural network weights is used for selecting the most relevant thermocouple inputs. Preliminary results show the strategy to be capable of holding the error to within 0.020 mm in the Z direction and 0.002° in the θX direction. Similar results are obtained during a cutting test.