Synthesis of the CNC machine control algorithms optimized with respect to the contour error using the genetic algorithm

Abstract

Traditionally control systems of CNC machines are built using the positional system structure, e.g. the uncoupled system (US) tuned on the symmetrical optimum. These systems have a relatively small contouring quality factor–the ratio of the contour speed to the contour error. To increase the quality factor of such systems, we proposed two new algorithms: the US tuned on the contour error optimum (US-C) and the neural coordinated control algorithm (NCCA). Both algorithms were tuned on a non-convex criterion using the genetic algorithm which allowed us to reduce the tuning time and acquire better optimal values of the criterion in comparison to the gradient descent method. The experiments were conducted using a mathematical model of a CNC machine based on real servo motors and lead screws. The results of the experiments showed that the usage of the proposed algorithms led to a significant increase in the quality factor, which has a positive effect on the overall productivity of the CNC machine. Furthermore, the NCCA use available resources better than the other tested algorithms which can be seen in the experiments with a lower initial contour error, a lower radius of the circle, and a higher contour speed. Besides, the US-C can maintain the contour speed better than other tested algorithms.