Accuracy enhancement of multi-axis CNC machines through on-line neurocompensation

摘要

This research is devoted to one of the most fundamental problems in precision engineering: machine tool accuracy. The paper presents a new approach designed to improve the accuracy of multi-axis CNC machines through software compensation of geometric, thermal and dynamic errors. Based on a multi-sensor monitoring system, the proposed compensation scheme is built to ensure error prediction. Four steps are required to develop and implement this system: (i) measurement of individual error components along each axis using a laser interferometer system, (ii) sensor integration via an artificial neural network model for on-line error estimation, (iii) synthesis of the total error into a three-dimensional error form using a simplified kinematic model and finally (iv) error compensation. Implemented on a turning center, the neurocompensation approach has improved machine accuracy by reducing the maximum error without compensation from 70 μm without compensation to less than 4μm.