Local shortwave irregularities detection for high-speed railway bridges using wavelet energy features of train-induced accelerations

Local shortwave irregularities, a prevalent form of track damage, intensify train-bridge interactions, potentially compromising the safety and comfort of high-speed trains while shortening the service life of trains, tracks, and infrastructure. Train-induced accelerations of bridges contain rich time-frequency vibration characteristics that can serve as critical indicators of such irregularities. This paper proposes a local shortwave irregularities detection method based on wavelet energy features, which quantifies the dynamic effects of these irregularities on train-induced accelerations. An objective function based on the ratio of intra-class feature variance to inter-class differences is developed, and the genetic algorithm is employed to optimize and select the optimal frequency interval. Within this interval, the extracted features exhibit high sensitivity to local shortwave irregularities while remaining robust against variations in train speed and axle load. A case study on a high-speed railway box girder bridge validates the effectiveness of the proposed algorithm through train-track-bridge interaction dynamic simulations and dynamic load tests. The results demonstrate that the wavelet energy feature-driven method provides a quantitative and interpretable description of local shortwave irregularity-induced vibrations, thus enhancing track maintenance predictability.