Towards intelligent control system for computer numerical control machines

Abstract

Advances in deep learning have led to impressive results in recent years. The new technologies such as convolutional neural networks, reinforcement learning and generative adversarial networks have shown a real promise for industrial and real-life applications. In this paper, the results of the experimental research on designing, training and implementation of the intelligent control system for the computer numerical control (CNC) machine were presented. The results indicate that using the generative adversarial technique in conjunction with reinforcement learning is possible to design and train the control systems for the machine tools. Building intelligent models in the absence of large datasets of labelled data is a crucial task. One of the key points of this experimental study is the training of a model of the control system using a set of unmarked data. This is achieved by using a reinforcement learning technique. A designed model can be deployed on the physical machine tools like a computer numerical control machine. At the presented research the laser engraver CNC machine is used. In this paper, the architecture of the computer intelligent control system for the laser engraver and the process of its training are described. The proposed model can be applied to different types of CNC machines