[152] Looseness detection system of bolted joints using a VMD-based nonlinear transformation approach with deep residual network
작성자 관리자
날짜 2025-01-15 15:19:42
조회수 35
D. Y. Kim, M. J. Kim, C. I. Kim, and G. H. Yoon. Measurement Science and Technology (2025).
This study develops a looseness detection system of bolted joints using a Variational Mode Decomposition (VMD)-based Nonlinear Transformation (NT) approach combined with a deep residual neural network. Bolted structures are susceptible to loosening due to environmental factors and various vibrations, which can significantly influence a structure's stability. Moreover, detecting the looseness of bolted joints can pose a challenge in intricate and large-scale structures. To detect the looseness accurately, the present study proposes a novel detection method that utilizes an NT through VMD applied to transverse vibrational modes. The VMD method decomposes transverse vibrational modes into Intrinsic Mode Functions (IMFs), selectively extracting the relevant modes that carry meaningful information. The NT method is applied as a method of scaling and shifting between the extracted signals. Image-based spectrograms are generated using the difference between the transformed signals and the loosenesses are detected in the deep residual network. To verify the present method, several plates with bolted joints are considered.