pp. 5031-5040
S&M3848 Research Paper of Special Issue https://doi.org/10.18494/SAM5200 Published: November 29, 2024 Efficient Gear Monitoring: Applying Deep Learning Models with Symmetrized Dot Pattern and Discrete Wavelet Transform to Detect Defects in Spur Gears [PDF] Shu-Hsien Huang, Ting-En Wu, and Chia-Hung Lai (Received June 25, 2024; Accepted October 29, 2024) Keywords: symmetrized dot pattern (SDP), discrete wavelet transform (DWT), convolutional neural network (CNN), deep neural network (DNN), gear detection
Gears are frequently used in transmission components, and essential information regarding gear transmission can be effectively analyzed by performing vibration detection. Therefore, in this research, we proposed a method of using deep learning models to establish an effective defect detection system for spur gears. In this system, data are collected and transformed using the symmetrized dot pattern (SDP) and discrete wavelet transform (DWT) techniques to detect defects in spur gears. The results of this study revealed that convolutional neural network models and deep neural network models can perform SDP detection at accuracy levels of 99% and 96%, respectively. Therefore, SDP and DWT are suitable for detecting defects in spur gears.
Corresponding author: Chia-Hung LaiThis work is licensed under a Creative Commons Attribution 4.0 International License. Cite this article Shu-Hsien Huang, Ting-En Wu, and Chia-Hung Lai, Efficient Gear Monitoring: Applying Deep Learning Models with Symmetrized Dot Pattern and Discrete Wavelet Transform to Detect Defects in Spur Gears, Sens. Mater., Vol. 36, No. 11, 2024, p. 5031-5040. |