Domain Adaptation Technique for Motion-based Prognostic and Health Management on Robot Arm  
Author Hsuan-Wen Lu


Co-Author(s) Ting-Yuan Song; Yu-Ling Cheng; Ling-Chieh Kung; Chia-Yen Lee


Abstract There are several strategies which had been used in prognostic and health management (PHM), from the typical run-to-failure maintenance (also known as reactive maintenance, RM), preventive maintenance (PM), conditional-based maintenance (CBM), to predictive maintenance (PdM). The PdM has become the most popular approach in the manufacturing industry. The remaining useful life (RUL) prediction is critical part of PdM, based on the health index (HI) construction. According to the collection of the equipment signals, such as vibration signals and acoustic emission signals, the health states of machinery can be measured and detected in real time. To further investigate the degradation processes of machinery, several features such as statistics, root mean squared, kurtosis, skewness, etc. are extracted from the monitoring signals in order to contract the His. However, different equipment could show a diverse nature; that is, one specific feature could be the HI for one machine, but this feature may not be successfully used for the other one even in the same machine type.
This study focuses on the signals which were captured from the machine’s robot arm in the TFT-LCD industry. The characteristic of the robot arm has a long useful life with an expensive repair costs, including the procurement cost, capacity loss, human resource and so on. The HI was constructed to estimate the health status of the arm then we can predict the RUL of this arm using machine learning technique. Furthermore, we use the domain adaptive technique to generalize on the same type of robot arm with different working conditions.


Keywords Domain adaptation, remaining useful life, prognostic and health management (PHM), transfer learning
    Article #:  RQD27-11

Proceedings of 27th ISSAT International Conference on Reliability & Quality in Design
Virtual Event

August 4-6, 2022