Automated polarisation-optical method for diagnostics of railway rails surface defects

This paper presents a novel polarisation-optical method for the automated inspection of surface defects in railway rails. The proposed approach is based on the detection of changes in the polarisation state of reflected laser radiation, caused by local anomalies such as cracks, residual stress zones, and surface contamination. A differential signal is generated by separating orthogonal polarisation components using a polarising beam splitter, enabling high sensitivity to surface irregularities while suppressing common-mode noise. To improve the interpretability of the acquired signal, a multi-stage digital processing pipeline is employed. It includes moving average and Gaussian filtering, threshold-based segmentation, and frequency-domain analysis using both discrete Fourier and continuous wavelet transforms. The method was validated through a set of simulated signals, imitating typical rail defects in noisy conditions. The results demonstrate reliable detection and localisation of structural anomalies, with a clear distinction between sharp discontinuities and wide modulations caused by tension. Due to its compatibility with automated processing frameworks, the method is well suited for integration with modern diagnostic platforms, including mobile rail inspection systems. This makes it a promising candidate for predictive maintenance strategies and digital transformation of railway infrastructure monitoring.