工具機智慧監測技術之發展與應用

隨著智慧製造與工業4.0 發展，工具機及其關鍵零組件之健康管理，已成為提升加工品質、降低設 備停機風險與維護成本的重要課題。傳統工具機維護多依據固定時間或累積運轉時數進行預防性保養， 較難即時反映主軸與軸承之實際劣化狀態。近年來，感測器、資料分析與人工智慧技術快速進步，使工 具機狀態監測逐漸由定期維護，轉向以即時量測與資料驅動分析為基礎之智慧化預知保養。本文首先介 紹兩種工具機主軸與軸承狀態監測技術，包含以加工製程監測資料結合非監督式機器學習，用於分析主 軸故障與製程事件關聯之方法，以及以數位－實體雙驅動多尺度轉移技術，解決軸承故障樣本不足問題 之診斷方法。接著，本文以多物理量智慧化量測模組為案例，說明預壓、振動與溫度等訊號之整合方式， 以及主軸軸承健康度分級模型之建立方法，並探討其於主軸跑合、軸承狀態辨識與早期異常預警之應用。 透過上述內容，本文期望提供工具機主軸狀態監測技術開發與產業導入之參考。

With the development of smart manufacturing and Industry 4.0, health management of machine tools and their key components has become an important issue for improving machining quality, reducing unexpected downtime, and lowering maintenance costs. Conventional machine tool maintenance is generally performed according to fixed time intervals or accumulated operating hours, making it difficult to reflect the actual degradation status of spindles and bearings in real time. In recent years, advances in sensors, data analytics, and artificial intelligence have enabled machine tool condition monitoring to shift from scheduled maintenance toward intelligent predictive maintenance based on real-time measurement and data-driven analysis. This article first introduces two condition monitoring approaches for machine tool spindles and bearings: one method applies in-process monitoring data combined with unsupervised machine learning to analyze the relationship between spindle faults and machining events; the other uses a digital–physical dual-driven multi-scale transfer method to address the problem of insufficient bearing fault samples. Next, an intelligent multi-physical measurement module is presented as a case study, illustrating how preload, vibration, and temperature signals are integrated to establish a health classification model for spindle bearings, with potential applications in spindle running-in, bearing condition identification, and early anomaly warning.