티타늄 합금 절삭 가공 시 딥러닝 기반 실시간 공구 상태 예측 시스템 개발

Deep Learning–Based Real-Time Tool State Prediction System for Titanium Alloy Machining

초록

Titanium alloys are widely used in aerospace and defense components; however, their low thermal conductivity accelerates tool wear and increases the risk of tool breakage, highlighting the need for real-time monitoring. This study proposes an AI-based tool state prediction system that integrates vibration sensing, deep learning, and real-time process visualization. Acceleration signals were collected using DAQ equipment, and multiple features were extracted to capture wearrelated dynamic changes. A two-stage learning structure was implemented: a classification model first separated cutting and non-cutting intervals to reduce noise, and a 1D CNN–Bi-LSTM regression model then predicted flank wear using only the cutting-state data. The workflow was implemented using a combined LabVIEW–Python architecture for real-time monitoring. RMS and variance were the most effective wear indicators, and the proposed model achieved high predictive accuracy, demonstrating the feasibility of real-time tool wear prediction in titanium alloy milling.

키워드

Deep learning; Cutting process; Real-time; Titanium alloy; Tool state prediction
제목
티타늄 합금 절삭 가공 시 딥러닝 기반 실시간 공구 상태 예측 시스템 개발
제목 (타언어)
Deep Learning–Based Real-Time Tool State Prediction System for Titanium Alloy Machining
저자
김휘협; 윤왕호; 김철민; 김미루
발행일
2026-10
유형
Y
저널명
한국생산제조학회지
권
35
호
5
페이지
439 ~ 447