Abstract
Tool breakage monitoring is a critical function in intelligent machine tool monitoring. Sudden failures such as chipping and complete breakage may rapidly degrade surface quality, cause workpiece scrap, and even induce secondary damage to machine components. To address the contradiction between response speed and robustness in conventional methods, this paper presents a fast-response online tool breakage monitoring system for engineering applications. The proposed system adopts a closed-loop architecture integrating sensing, decision-making, alarm output, and machine interlocking. Multi-source signals, especially vibration and cutting-related signals, are processed through a lightweight low-latency pipeline including filtering, denoising, and envelope enhancement. A collaborative decision strategy combining fixed thresholds and dynamic thresholds is developed, where fixed thresholds are used for the rapid interception of severe abnormalities such as complete tool breakage and large chipping, while dynamic thresholds adapt to local signal statistics for the detection of slight abnormal conditions. To verify the real-time performance, an impact-hammer-based experiment was conducted, and an oscilloscope was used to synchronously record the abnormal excitation instant and the alarm output instant. Experimental results show that the front-end sensing chain responds within approximately 51.2 μs, and the end-to-end alarm output response time is about 1.362 ms, which satisfies the engineering requirement of less than 10 ms. After abnormality detection, the system can reliably output a high-level control signal to trigger the feed-hold protection of the machine tool. The results demonstrate that the proposed system provides millisecond-level online protection capability and strong engineering feasibility for intelligent machine tool monitoring.