TY - GEN
T1 - Total precision inspection of machine tools with virtual metrology
AU - Tieng, Hao
AU - Yang, Haw Ching
AU - Cheng, Fan Tien
PY - 2015/10/7
Y1 - 2015/10/7
N2 - The purpose of this study is trying to apply VM [1][2] for measuring machining precision of machine tools. As shown in the left portion of Fig. 1, the AV M server [3] needs process data, which contain sensor data and machining parameters, as inputs for predicting machining accuracies. However, most machine tools do not possess sensors for providing raw process data such as vibration, current, etc. Furthermore, machining operations usually possess severe vibrations and loud noises. It makes the raw data collected from sensors attached to machine tools with low signal-to-noise (S/N) ratios, thereby affecting the prediction accuracy of VM. Even though the AVM has been an effective way to conduct workpiece measurements in high-tech industries [1], there exist challenges when applying the AVM to machining industry. Besides embedding essential sensors on the machine tool effectively, the challenges are 1) Segmentation: to accurately segment essential part of the raw process data from the original NC file, 2) Data Cleaning: to effectively handle raw process/sensor data with low S/N ratios, and 3) Feature Extraction: to properly extract significant features from the segmented raw process data. This work proposes a novel GED-plus-AVM (GAVM) system as depicted in Fig.1 to resolve all the challenges mentioned above. These challenges are judiciously addressed and successfully resolved in this paper.
AB - The purpose of this study is trying to apply VM [1][2] for measuring machining precision of machine tools. As shown in the left portion of Fig. 1, the AV M server [3] needs process data, which contain sensor data and machining parameters, as inputs for predicting machining accuracies. However, most machine tools do not possess sensors for providing raw process data such as vibration, current, etc. Furthermore, machining operations usually possess severe vibrations and loud noises. It makes the raw data collected from sensors attached to machine tools with low signal-to-noise (S/N) ratios, thereby affecting the prediction accuracy of VM. Even though the AVM has been an effective way to conduct workpiece measurements in high-tech industries [1], there exist challenges when applying the AVM to machining industry. Besides embedding essential sensors on the machine tool effectively, the challenges are 1) Segmentation: to accurately segment essential part of the raw process data from the original NC file, 2) Data Cleaning: to effectively handle raw process/sensor data with low S/N ratios, and 3) Feature Extraction: to properly extract significant features from the segmented raw process data. This work proposes a novel GED-plus-AVM (GAVM) system as depicted in Fig.1 to resolve all the challenges mentioned above. These challenges are judiciously addressed and successfully resolved in this paper.
UR - http://www.scopus.com/inward/record.url?scp=84952763515&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84952763515&partnerID=8YFLogxK
U2 - 10.1109/CoASE.2015.7294301
DO - 10.1109/CoASE.2015.7294301
M3 - Conference contribution
AN - SCOPUS:84952763515
T3 - IEEE International Conference on Automation Science and Engineering
SP - 1446
EP - 1447
BT - 2015 IEEE Conference on Automation Science and Engineering
PB - IEEE Computer Society
T2 - 11th IEEE International Conference on Automation Science and Engineering, CASE 2015
Y2 - 24 August 2015 through 28 August 2015
ER -