A multi-point excitation and spectral clustering-weighted cross-spectral correlation method for Bogie frame damage identification

The rapid propagation of certain cracks in the spring sleeve of the metro bogie frame, once they reach a detectable size, poses a significant risk. If missed during a visual inspection, it can result in structural failure prior to the next maintenance check. To mitigate this safety hazard, this study targets cracks in the spring sleeve and proposes a novel crack detection method based on single-point sequential multi-location excitation, tailored for low-level maintenance routines. This method leverages a spectral clustering weighted cross-spectral correlation algorithm with a dynamic scale adaptation mechanism (SCWCC-DSAM) to achieve precise and efficient crack identification. The SCWCC-DSAM method is validated through single-point excitation tests. By partitioning and weighting frequency response functions (FRFs), the method significantly improves identification accuracy for medium and large cracks, outperforming other FRF-based methods such as cross-spectral correlation. Finally, multi-position excitation tests based on the proposed method are conducted. By combining a statistical threshold (determined as 0.051 in this study) with the damage monitoring matrix, cracks smaller than the limit size are successfully identified and localized under limited sensor conditions, achieving regional localization confined to the sleeve region. This method provides important technical support for the safe operation of bogie frames. It enables early detection of cracks in critical areas during routine maintenance, reduces the risk of sudden fracture, and underpins the safe and reliable operation of metro vehicles.