貝氏分類應用於研磨之完工品質聚類分佈

本文研究平面磨床振動信號，監控研磨加工過程中之機台主軸振動對於完工品質狀態分佈。根據統計參數對完工成品之影響分析，探討在不同參數下所產生之影響，進行工況調整檢測，對異常工況提出警告，以便加工人員採取應變措施，消除異常製程恢復製程穩定。振動信號採樣為加速規模組，建置於平面磨床主軸上，擷取完整的研磨加工振動信號，以統計參數計算萃取出分佈狀態之特徵值，如波形、峰值、裕度、峭度及歪度，將特徵值建立於訓練樣本中，計算其協方差矩陣作為分類工具，由貝氏理論作為分類基準，進行完工品質狀態分類，並建立完工品質狀態之訓練樣本於資料庫中。資料前處理部分，加速規之濾波功能與演算法已建置於系統模組中，以貝氏分類模組作為分類依據建立於LabVIEW人機介面上，將其研磨完工狀態利用人機界面顯示，便於使用者操作及了解研磨完工品質狀態。研磨加工實驗結果可區分出不同狀態下之完工品質，如鏡面、霧面、顫紋、刮傷及燒傷，利用此研磨完工狀態分類達到品質監控、穩定製程、提升製程良率及可能性。

關鍵字

Bayes分類；平面磨床；振動信號；品質監測

並列摘要

This article surface grinder vibration signal, monitoring the distribution of machine spindle vibration for the quality of the state of completion of the grinding process. Statistical parameters of the completion of the finished product analysis, the effects of different parameters and conditions to adjust the detection and warning of abnormal conditions, so that the processing of personnel to take contingency measures to eliminate abnormal process to restore the stability of the process. Vibration signal is sampled to accelerate scale group, built on a surface grinder spin -dle retrieve the complete grinding vibration signal statistical parameters extracted from the characteristics of the distribution of values, such as waveform, peak margin, kurtosis and skewness eigenvalue establish training samples, calculate it covariance matrix as classification tools, as by the theory of Bayesian classification basis, completion quality state classification, and the establishment of the completion of the quality status of the training samples in the database. Information before processing part, Accelerometers filtering algorithm has been built on the system module, Bayesian classification module as a classification basis is based on the Lab VIEW front panel, the part of it grinding completion state HMI display, user will be friendly and understand for the grinding completion of the quality status. Grinding experimental results can distinguish the quality of the different state of completion, such as mirror、matte、flutter pattern、scrapes and burns, use this grinding completion status classification to achieve will be quality control, and stable process、to improve the yield rate and possibility.

並列關鍵字

quality monitoring. ； vibration signal ； surface grinder ； Bayes classifier

參考文獻

[9] 王俊傑，旋轉機械故障診斷方法之探討，中原大學博士論文，2007。

[1] Samanta, B. and Al-Balushi, K. R., “Artificial Neural Network Based Fault Diagnostics of Rolling Element Bearings Using Time-Domain Features,” Mechanical Systems and Signal Processing, Vol. 17, No. 2, pp.317-328, 2003.

[2] S. Dey, J.A. Stori “A Bayesian network approach to root cause diagnosis of process variations” Department of Mechanical and Industrial Engineering, IL 61801, USA, 2003.

[4] Rafl'aella Settimi, and Jim Q. Smith, “on the geometry of bayesian graphical models with hidden variables,” University of Warwick, Coventry CV4 7AL (U.K.) ,1997.

[6] Romessis, C. and Mathioudakis, K., “Bayesian Network Approach for Gas Path Fault Diagnosis,” Journal of Engineering for Gas Turbines and Power, Vol. 128, No. 1, pp. 64-72, 2006.

國際替代計量

貝氏分類應用於研磨之完工品質聚類分佈

未授權

主題瀏覽