Design and Development of a Real-Time AI-Based Tool Failure Prediction System for Machining Difficult-to-Cut Materials

Cutting processes are essential in manufacturing metal components, particularly in aerospace applications. For difficult-to-cut materials such as titanium alloys, tool condition significantly impacts machining quality, making real-time failure prediction critical. While many studies have applied artificial intelligence (AI) to analyze sensor data for predictive maintenance, real-time AI implementation during cutting remains limited. This study presents a real-time tool failure detection system capable of executing an AI model during machining. Titanium alloy was machined using an end mill, and an acceleration sensor was mounted on the machine tool for condition monitoring. Data acquisition, preprocessing, and transmission to the AI model were handled via LabVIEW. The AI model classified the tool condition into four states and was integrated with the monitoring system through TCP/IP communication for real-time prediction. The system was validated through cutting experiments, demonstrating its effectiveness for real-time tool condition monitoring in high-performance machining environments.