Bidirectional Long Short-Term Memory Approach for Infrastructure Health Monitoring Using On-Board Vibration Response

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Deep learning estimates railway track stiffness from drive-by vibration response data, highlighting feature extraction and bidirectional analysis.

The LSTM-BiLSTM model accurately estimates railway track stiffness by utilizing drive-by vibration response signals, reducing noise impact.
Maximum mean absolute percentage errors were reported at 1.7% for railpad stiffness and 0.7% for ballast stiffness, indicating strong performance.
Framing of vibration signals was employed to enhance monitoring resolution at the beam level, improving condition estimation accuracy.
The methodology shows potential for automated infrastructure health assessments across various bridge and railway conditions.

Samani et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

Deep learning estimates railway track stiffness from drive-by vibration response data, highlighting feature extraction and bidirectional analysis.