Data-driven modeling of high-speed maglev track irregularity

Ensuring the stability of high-speed maglev trains hinges on track smoothness, which is influenced by track irregularities that act as key excitations for train vibrations. These irregularities, stemming from various factors including track design and environmental conditions, are unpredictable and dynamic. Current models often fail to accurately represent these irregularities, leading to unreliable dynamic analyses. This paper introduces a non-stationary, non-Gaussian stochastic process model, enhanced with Iterative Amplitude Adjusted Fourier Transform (IAAFT) and Time-series Generative Adversarial Network (TimeGAN) algorithms, to more accurately simulate track irregularities. The model’s ability to generate independent, high-fidelity data supports improved design, operation, and maintenance of maglev systems.