기계 학습 기법을 이용한 밀링 공정 중 엔드밀 마모 진단 시스템

초록

The quality of a product significantly varies depending upon the wear condition of a tool during a milling process. In industrial sites, a change in the surface condition of the product or the sound of processing is detected, and the tool is visually inspected to determine the wear state. In this study, a technique was developed for wear state identification of tools using audio data to prevent the errors caused due to visual inspection. The audio data was recorded during the milling process, and the data dimensionality reduction was performed using principal component analysis (PCA) and partial least squares (PLS). The data were classified using kernel support vector machine (SVM) by applying various functions.

키워드