Global sensitivity analysis of parameters for medium and low-speed maglev vehicle-bridge coupling system based on an improved Sobol method
To identify the key parameters of a medium and low-speed maglev vehicle–bridge coupling system, a quantitative parameter sensitivity analysis is essential. To address the heterogeneous parameter dimensions, the difficulty of globally evaluating time-varying responses, and the prohibitive cost of applying the Sobol method directly to the coupling model, this paper proposes an improved Sobol global sensitivity analysis method integrating an error integral performance indicator with a Levenberg–Marquardt Back-Propagation (LMBP) neural network surrogate. The time-varying suspension-gap response is first scalarized by the error integral indicator, which eliminates dimensional differences among the parameters and condenses the global dynamic behaviour. The surrogate then establishes the mapping from the system parameters to the indicator, avoiding repeated time-consuming simulations of the original model. Based on the surrogate, Sobol-sequence sampling is performed and the parameter sensitivities are computed, revealing the interaction mechanisms of the parameters on the dynamic behaviour and providing a theoretical basis for engineering parameter optimization.
Authors
- Fei Zhou (ORCID: https://orcid.org/0009-0005-5929-5290)
- Zhiqiang Long
- Xiaolong Li
- Zhenghao Gu (ORCID: https://orcid.org/0009-0006-6742-597X)
Institutions
- National University of Defense Technology (CN)
Publication Details
- Journal
- Journal of Vibration and Control
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1177/10775463261508028
- Primary Topic
- Railway Engineering and Dynamics
- Type
- article
- Field-Weighted Citation Impact
- 0.00