A Study on Artificial Neural Network Learning Using Decimal Data of CNC Cutting Conditions for Aluminum

Abstract

The processing conditions of computer numerical control (CNC) equipment are learned through practical experience over many years. However, it is currently difficult to train professional experts in the field of machining because of the “demographic cliff” phenomenon. To solve this issue, it is necessary to study how to transfer the experience of current experts to new people who desire to use CNC equipment. An “expert system for CNC equipment” is proposed using artificial neural networks (ANNs) and machine-learning data is made available in decimal rather than binary so that mechanical engineers can easily access artificial intelligence. Assuming that an aluminum alloy (AL6063) is machined by a beginner using a lathe, when the tensile strength, cutting speed, and desired surface roughness are input into the ANN, the rotation speed (revolutions per minute) of the spindle and the feed rate are output by the proposed expert system for CNC equipment setting. This study demonstrates that decimal data about aluminum can be used for learning by an ANN, and it is judged that learning by using decimal data is possible for various materials and in various processing situations.