Real-time Positioning Error Compensation for a Turning Machine Using Neural Network

Abstract The dimensional accuracy of machined parts mainly depends upon the positioning accuracy of machine tool slides. An adaptive and active approach of compensating positioning error has been demonstrated in this paper, in order to bring down the positioning error of a high precision slide of a CNC turning machine in sub micrometer range . A module has been developed for real time computation and compensation of positioning error of an axis through a open architecture motion controller, using back propagation feed forward neural network. The training of neural network has been carried by mapping the positioning error using a laser interferometer. Key network parameters of the trained neural network has been extracted and used in a ‘C’ program for real time computation and compensation of the positioning error based on the position of the axis, by updating tool offset using global variable. The positioning error of the slide has been reduced to sub micrometer range after implementation of the module in the motion controller the high precision turning machine.