Modeling uncertainties of vehicle-track coupled dynamic systems

In this paper, the nonparametric models of the vehicle and track subsystems are established separately by the nonparametric probabilistic method, in which modeling uncertainties are represented by random matrices and the dispersion level of each random matrix is controlled by a corresponding dispersion parameter. The stochastic equations of motion of the vehicle and track subsystems are obtained respectively and coupled by the wheel-rail dynamic interaction model, in which nonlinear wheel-rail contact geometry relation and wheel-rail detachment can be considered. Random responses are obtained with the prediction-based iterative method in the framework of Monte Carlo simulation (MCS). The sensitivities of the random responses to the uncertainties of coefficient matrices of vehicle and track subsystems are analyzed and the confidence intervals obtained by the nonparametric and parametric models are compared. The results show that the lateral acceleration of the car body on a straight track is most sensitive to the vehicle's mass matrix, while the wheel derailment coefficient is most sensitive to the vehicle's stiffness matrix at curve negotiation. The confidence intervals obtained by the two probabilistic models with the same dispersion level are close to each other, verifying the validity of the nonparametric model of vehicle-track coupled systems. Communicated by Corina Sandu.

Automated summary

In this paper, nonparametric models of vehicle and track subsystems are established separately using a probabilistic method. The sensitivity of random responses to uncertainties in coefficient matrices is analyzed, and confidence intervals from nonparametric and parametric models are compared. The results validate the effectiveness of the nonparametric model for vehicle-track coupled systems.