Compensation for Positioning Error of CNC Machine Tools Using Neural Network. Compensation for Multivariant Positioning Error.

Abstract

In this paper, it is shown that traveling length, feed speed, proportional gain and other factors affect the positioning error of CNC machine tools, and that the positioning error can be regarded qualitatively as a multivariant function in high-precision positioning. The positioning error due to these factors is difficult to be solved by traditional compensation methods because the relationship between the positioning error and the machine environment cannot be mathematically expressed. Thus a compensation method using a neural network was proposed in the study. The neural network was used to learn the relationships between the positioning error and the machine environments, and the Back Propagation algorithm was used as the learning algorithm. Through experiments and computer simulation, it is shown that with the aid of a neural network, the nonlinear problems of the positioning error of CNC machine tools can be solved easily and with high-precision.