本研究目的為搭配智能化對稱雙主軸研磨機開發一套監控系統,雙主軸研磨機是用來製造碳化鎢組成的LED探針,此監控系統包含三個部份: (1)藉由架設在高速主軸上的類比加速規擷取振動訊號。 (2)資料擷取卡模組會取樣類比訊號並將訊號轉換為數位訊號儲存。 (3)透過Visual C#語言可以將擷取出的數位訊號利用演算法進行分析。 透過監控系統可以達到以下幾點: 1.監控系統可以偵測出聚晶鑽石磨輪上的鈍化、填塞、磨損現象。 2.根據振動訊號調整LED探針的製程參數,使雙主軸研磨機有更好的加工性能。 3.透過軸承的異常偵測演算法能夠提早偵測到主軸的異常發生,監控軸承會發現到主軸隨著運轉的時間增加造成加工性能衰退,這在預知保養是非常重要的一環。至今,旋轉機械的複雜程度已經日異月新,當機械發生異常時,操作者不可能即時透過肉眼找出機械的問題所在,因此發展可以即早偵測出異常且避免突然發生的故障的預知保養系統是一件重要的工作,本研究提出方均根、峭度、排序熵及支持向量資料描述法當作異常偵測的方法。 根據實驗結果證明本論文所開發監控系統的效率性及可行性。
The objective of this study is to develop a monitoring system for intellectualized symmetric high‐speed dual‐spindles grinding used for WC LED‐probe fabrications. This monitoring system consists of three parts: 1) vibration signals collected by analog accelerometers installed on the spindle; 2) the analog signals are sampled and converted into digital numeric values via data acquisition (DAQ) modules; 3) the resulting digital samples are then analyzed by using algorithms implemented by Visual C# language. Based on this monitoring system, the following tasks were performed: 1. The wear, clog and dulling in the PCD wheel can be detected out from the developed monitoring system. 2. The process parameters for LED‐probe fabrications are adjusted to obtain a better performance according to the vibration signals. 3. The anomaly of the spindle can be detected early by a novelty bearing anomaly detection algorithm. Monitoring the performance degradation of bearings in spindle continuously is necessary for predictive maintenance. Yet, the complexity of rotary machine increases dramatically and it is impossible for an operator to keeping an eye on the machine’s condition daily. Therefore, it is essential to develop a prognostics system that has the ability to detect abnormality early and prevent unexpected failure. In this study, a data-driven prognostic methodology based on RMS, Kurtosis, Permutation Entropy and support vector data description (SVDD) is proposed. The experimental results demonstrates the efficiency and feasibility of the developed monitoring system.