TY - GEN
T1 - Wear Particle Texture Analysis
AU - Laghari, Mohammad Shakeel
AU - Hassan, Ahmed
N1 - Funding Information:
ACKNOWLEDGMENT This work is supported by UAEU Program for Advanced Research under the Grant/Approval No: (31N321-UPAR -2-2017).
Publisher Copyright:
© 2019 IEEE.
PY - 2019/7
Y1 - 2019/7
N2 - Wear particle analysis is used for health monitoring of a machine to identify possible failure modes resulting from various machine components. One of the initial phases of analysis typically involves a microscopic examination of particles extracted from the lubrication system. Experts in the field (Tribologists) utilize this important information from these examinations to monitor the operation of the machine and ensure efficiency, safety, and economy of the operation. This extracted information has been characterized by six morphological features of particle shape, edge curvature, size, color, thickness ratio, and surface texture. These features are the building blocks in the creation of image analysis systems to supplement the wear judgments made by experts who use traditional methods of diagnostics. The paper investigates one of the six morphological features related to the surface texture classification of wear particles. Image processing procedures of line/edge detection and Artificial Neural Networks are used to recognize four texture types of smooth, pitted, striations, and rough. An accuracy classification rate of approximately 98.9% has been achieved and is shown by a confusion matrix.
AB - Wear particle analysis is used for health monitoring of a machine to identify possible failure modes resulting from various machine components. One of the initial phases of analysis typically involves a microscopic examination of particles extracted from the lubrication system. Experts in the field (Tribologists) utilize this important information from these examinations to monitor the operation of the machine and ensure efficiency, safety, and economy of the operation. This extracted information has been characterized by six morphological features of particle shape, edge curvature, size, color, thickness ratio, and surface texture. These features are the building blocks in the creation of image analysis systems to supplement the wear judgments made by experts who use traditional methods of diagnostics. The paper investigates one of the six morphological features related to the surface texture classification of wear particles. Image processing procedures of line/edge detection and Artificial Neural Networks are used to recognize four texture types of smooth, pitted, striations, and rough. An accuracy classification rate of approximately 98.9% has been achieved and is shown by a confusion matrix.
KW - Artificial Neural Networks
KW - image processing
KW - texture classification
KW - wear particles
UR - http://www.scopus.com/inward/record.url?scp=85078106227&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85078106227&partnerID=8YFLogxK
U2 - 10.1109/ICISPC.2019.8935804
DO - 10.1109/ICISPC.2019.8935804
M3 - Conference contribution
AN - SCOPUS:85078106227
T3 - 2019 3rd International Conference on Imaging, Signal Processing and Communication, ICISPC 2019
SP - 67
EP - 72
BT - 2019 3rd International Conference on Imaging, Signal Processing and Communication, ICISPC 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 3rd International Conference on Imaging, Signal Processing and Communication, ICISPC 2019
Y2 - 27 July 2019 through 29 July 2019
ER -