|
CSME 2025/04
Volume 46 No.2
:
121-130
|
|
Application of Multiclass Classification to Fault Diagnosis in Harmonic Drive
Tsung-Yu Yu a, Nan-Kai Hsieh b and Shiaw-Wu Chen b
aPh.D. Program of Electrical and Compunctions Engineering, Feng Chia University, Taiwan, R.O.C.; Department of Computer Science and Information Engineering, National Taichung University of Science and Technology, Taiwan, R.O.C.
bDepartment of Automatic Control Engineering, Feng Chia University, Taiwan, R.O.C.
|
Abstract:
The harmonic drive has become a critical component in robotic arms, enhancing their load capacity. Therefore, the harmonic drive's health status affects robotic arms' operational stability. This study focused on diagnosing anomalies in harmonic drives before equipment failure. By artificially creating five common types of faults in harmonic drives and collecting vibration signals with a three-axis accelerometer, this study trained and verified thediagnostic capabilities of numerous classification algorithms, namely the random forest, K-Nearest Neighbors (KNN), Support Vector Classification (SVC), and eXtreme Gradient Boosting (XGBoost) algorithms. In the experiments performed in this study, SVC and XGBoost exhibited excellent abilities in identifying harmonic drive faults and classifying potential fault causes. Thus, these algorithms can facilitate the adoption of immediate fault-preventionmeasures.
|
|
Keywords: Harmonic Drive, Prognostic and Health Management (PHM), Machine Learning, Multiclass Classification
|
|
|
*Corresponding author; e-mail:
|
©
2025
CSME , ISSN 0257-9731
|