Toward compressive sensing of track irregularities

The management of massive track irregularity data from extensive railway networks poses significant challenges in storage, transmission, and processing. Compressive sensing (CS), which enables sub-Nyquist sampling by leveraging signal sparsity, presents a potential solution. However, its application to track irregularities, which are characterized by broad bandwidth, stochasticity, and stringent requirements for high-fidelity reconstruction in both spatial and frequency domains, faces unique obstacles, and thus remains underexplored. This study systematically analyzes the applicability of most compatible sparse representation bases, including Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), and Discrete Fourier Transform (DFT), to track irregularity signals. Through sparsity analysis and multi-dimensional performance evaluation (e.g., NRMSE, R 2 , PAE, SPD), it is demonstrated that the DFT provides the most effective sparse representation. The DFT-based CS framework enables compression ratios of track irregularity data up to 50% while maintaining high fidelity. Nevertheless, DFT-based reconstruction involves higher computational cost than its DCT counterpart. Accordingly, DCT is a practical choice for efficiency-oriented applications, as validated by train-track dynamic simulations. This work establishes a practical and efficient CS-based solution for large-scale track irregularity data management.