銑削加工振動訊號與表面粗糙度之關聯性

The Correlation Between Vibration Signals and Workpiece Surface Roughness in Milling

洪沛志(Pei-Chih Hung)

指導教授 ： 李貫銘

國立臺灣大學/工學院/機械工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201802464

摘要

工具機隨著工業的發展越來越受到重視，尤其是在製造業更是核心所在，例如國防、航太產業、生醫產業、電子業、汽車工業等，都能看到其存在的重要性。而銑削加工中，表面粗糙度是加工品質的參考指標之一。在精加工中的表面粗糙度值越小越好。然而銑削加工過程的物理現象非常複雜，以理論公式推導表面粗糙度值，在實務上受到許多限制。若能以切削過程訊號輔助理論公式進行切削條件規劃，可省去反覆試驗的困擾。振動訊號是常見的切削訊號之一，其與表面粗糙度有高度的關聯性，但是振動訊號受機台特性、刀具幾何造型、工件材料性質與加工條件等的影響，不易直接由振動訊號推測表面粗糙度值。文獻中關於振動訊號直接對應表面粗糙度的研究很少。針對以上的問題，本研究擬觀測工具機上容易取得的加速規振動訊號作為銑削加工中表面粗糙度異常判斷的可行性。
本研究在機台之主軸與虎鉗上各安裝一顆三軸加速規，以及於工件上安裝一顆單軸加速規一同量測振動訊號，在實驗前加工參數設定以及刀具的選擇後，以實驗中擷取的振動訊號透過頻譜分析找出訊號特徵值，了解振動訊號與表面粗糙度的關聯性。接著以數學理論公式計算理論表面粗糙度，進行表面粗糙度實際值與理論值比較而分類，分為與理論值趨勢相同以及趨勢異常兩組。最後透過類神經網路建立演算法進行分類，並觀察其預測效果。在驗證實驗中，位於進給方向上的X方向訊號之模型分類結果存在較高準確率，其中在工件X向訊號於模型分類之準確率可高達92.9%，虎鉗X向訊號經模型分類後之準確率亦可達92.6%，而主軸X向訊號於模型分類後之準確率也有86.3%。本研究所建立之演算法可有效對於實際表面粗糙度進行正常、異常判斷。

關鍵字

振動訊號 ； 加速規 ； 表面粗糙度 ； 類神經網路

並列摘要

With Computer numerical control (CNC) machining centers getting more and more attention with the development of industry, we can see the importance of their existence especially in the manufacturing industry, like national defense, aerospace industry, biomedical industry, electronics industry and automobile industry. Surface roughness is one of the reference indicators of processing quality in milling. It’s better to get the smaller value of surface roughness in finishing processes. However, the physical phenomenon of milling process is complicated. Deriving the surface roughness value by theoretical formula is subject to many limitations in practice. The trouble of the repeated test can be omitted as the operating parameters are carried out with the signal-assisted theoretical formula of the cutting process. Vibration signal is one of the common signals in cutting, which has a high correlation with surface roughness. However, the vibration signal is affected by the dynamic characteristics of machining centers, tool geometry, material properties of workpiece and operating parameters. It is difficult to directly infer surface roughness from vibration signal. There is little research investigating the direct relationship between vibration signals and surface roughness. In view of the above problems, this study measures the vibration signals from accelerometers, investigating the feasibility of the abnormality of the surface roughness in milling.
In the experiment, 3-axis accelerometers were mounted on the spindle and the vise in the machining center, and a single-axis accelerometer was mounted on the workpiece. They were used to measure the vibration signal in milling. First, after operating parameters and tools were selected, we extracted vibration signals from milling, found out the features by spectral analysis method, and to know the correlation between vibration signals and surface roughness. Secondly, the actual surface roughness of the experiments was compared with the theoretical surface roughness calculated by the mathematical formula. Then the signals were classified into two labels: normal and abnormal conditions. Finally, the neural networks were established to classify the experimental signals and to predict the conditions of the actual surface roughness. The validation results showed that the higher accuracy was achieved with vibration signal in X-direction, which is the feed direction. The accuracies of X-direction vibration signal from the workpiece and the vise were 92.9% and 92.6% respectively. Besides, the accuracy was 86.3% from the spindle signal in X-direction. The neural networks can effectively classify the conditions of actual surface roughness.

並列關鍵字

Vibration Signals ； Accelerometer ； Surface Roughness ； Neural Network

參考文獻

[1] J. Corbett, R. A. McKeown, G. N. Peggs, and R. Whatmore, "Nanotechnology: International developments and emerging products," CIRP Annals - Manufacturing Technology, 49(2), pp. 523-545, 2000.

[2] R. Teti, K. Jemielniak, G. O'Donnell, and D. Dornfeld, "Advanced monitoring of machining operations," CIRP Annals - Manufacturing Technology, 59(2), pp. 717-739, 2010.

[3] M. C. Shaw, Metal cutting principles, 2nd ed., Oxford University Press, New York, 2005.

[4] N. Ambhore, D. Kamble, S. Chinchanikar, and V. Wayal, "Tool Condition Monitoring System: A Review," Materials Today: Proceedings, 2(4), pp. 3419-3428, 2015.

[5] M. Malekian, S. S. Park, and M. B. G. Jun, "Tool wear monitoring of micro-milling operations," Journal of Materials Processing Technology, 209(10), pp. 4903-4914, 2009.