Model-driven intelligent graded evaluation of high-speed maglev track irregularity and its application validation

Abstract

Track irregularities in high-speed maglev systems are critical factors influencing levitation stability and ride quality. Conventional evaluation methods struggle to simultaneously account for frequency-domain characteristics and engineering randomness, limiting the realization of intelligent perception and automated, scientifically grounded assessment of track conditions. To fill this gap, this study proposes a modeling and graded evaluation framework for track irregularities based on multi-category deviation superposition. A hierarchical superposition model is established to represent the coupled deformation of track girders and stator cores, and a random parameter assignment strategy is introduced to simulate geometric fluctuations under multi-source uncertainties. On this basis, a five-level evaluation criterion is developed using PSD confidence intervals, forming a probabilistic grading system suitable for automated track state inspection. The neighborhood uniform sampling Bootstrap approach is employed to reconstruct PSD statistical intervals from small-sample field measurements, confirming the consistency of the model and the rationality of the evaluation boundaries. Application studies demonstrate that the method can automatically and accurately identify irregularity degradation across different wavelengths on the Shanghai Maglev Line, and its comparative evaluation with the German TVE test line highlights strong cross-line applicability. The proposed track state intelligent perception and graded evaluation framework provides a scientific basis for managing irregularities and suppressing vibrations in high-speed maglev systems, supporting stable operation and improved ride quality under demanding dynamic conditions.