應用影像處理技術建立刀具剩餘壽命預測系統

The Prediction System for Tool's Remaining Useful Life Based on The Image Processing Techniques

陳俊宏(Jun-Hong Chen)

指導教授 ： 江佩如

國立中正大學/工學院/機械工程學系/碩士(2014年)

摘要

在改善自動化的加工過程中，刀具磨損狀態監測和剩餘使用壽命預測是非常重要。生產過程中刀具的磨損，將會造成一些負面的影響，如停機時間增加，生產效率降低，甚至會提高安全風險。為了避免上述問題產生，就必須得知在生產過程中刀具的剩餘壽命。因此，本研究提出一種以影像感測技術與系統鑑別理論為基礎的方法，將其應用於剩餘壽命的預測。本論文以車床做為研究對象，拍攝車刀表面影像作為監測的訊號，藉由影像處理技術分析車刀影像，並量測出車刀的刀腹磨損量。由於車刀的磨損量與加工時間的關係可以視為一個系統的時間響應，因此可以透過系統鑑別理論計算出磨損曲線的系統參數，最後藉由此系統即可以進行刀具剩餘壽命的預測，為了驗證此預測系統的準確性，本研究實際預測六把車刀在每次切削後的刀具剩餘壽命，將其與實際剩餘壽命進行比較，並計算預測誤差進入±10%後所有誤差的平均，可以得到一階系統中六組車刀的平均預測誤差為5.1%，二階系統中六組車刀的平均預測誤差為2.9%，而三階系統中六組車刀的平均誤差為3.6%。

關鍵字

剩餘壽命預測 ； 影像處理 ； 系統鑑別 ； 車刀磨損

並列摘要

In the process of automated production, to monitor tool wear and to predict the remaining useful life are very important. Tool worn occurs during production operation, which may result in several negative implications such as increase in downtime, low productivity and sometimes even cause safety risks. In order to avoid those previous problem, it is necessary to know the remaining useful life of tool. Therefore, this study provides a method for predicting the tools’ remaining useful life, based on the image processing techniques and system identification theory. This study did the research on the CNC cutting machine, and record the flank of cutting tool by photo. Then, we analyzed the tool's image and measure the flank wear by digital image technology. Due to the relationship of flank wear and processing time, it could be considered as the impulse response of system, then we can calculate the unknown system parameters of tool wear curved by system identification theory. Finally, the tool remaining useful life can be predicted by using the system model of tool wear curve. In order to prove the accuracy of this predicting system, this research tried to predict the remaining use of the six cutting tools, and tried to compare the remaining useful life, then calculate the average of the error under ±10%. By this research we could get an first-order system of the six tools which average prediction error was 5.1%. Also, the average error of the second-order system turning six groups was 2.9%. In addition, the average error of the third-order system turning six groups was 3.6%.

並列關鍵字

image processing ； system identification ； cutting tool wear ； remaining useful life prediction

參考文獻

[1] D. Snr and E. Dimla, "Sensor signals for tool-wear monitoring in metal cutting operations—a review of methods," International Journal of Machine Tools and Manufacture, vol. 40, pp. 1073-1098, 2000.

[3] S. Kurada and C. Bradley, "A review of machine vision sensors for tool condition monitoring," Computers in Industry, vol. 34, pp. 55-72, 1997.

[5] S. Kurada, C. Kurada, and Bradley, "A machine vision system for tool wear assessment," Tribology international, vol. 30, pp. 295-304, 1997.

[7] T. Pfeifer and L. Wiegers, "Reliable tool wear monitoring by optimized image and illumination control in machine vision," Measurement, vol. 28, pp. 209-218, 2000.

[9] J. Jurkovic, M. Korosec, and J. Kopac, "New approach in tool wear measuring technique using CCD vision system," International Journal of Machine Tools and Manufacture, vol. 45, pp. 1023-1030, 2005.