Sensitivity analysis and multi-objective optimization of the virtual track train parameters considering the coupling effect between vehicles

To improve vehicle stability, passenger comfort and road friendliness of the virtual track train (VTT) negotiating curves, a multi-parameter and multi-objective optimization platform combining the VTT dynamics model, Sobal sensitivity analysis, NSGA-II algorithm and k - epsilon optimal selection method is developed. Considering the mutual coupling between vehicles and taking random road roughness as excitation, the VTT dynamics model with 66-DOF is constructed and verified. Design variables and objective functions are defined, and then key design variables are obtained via Sobol sensitivity analysis. The NSGA-II algorithm is used to optimize the key design variables of the VTT, and the k - epsilon optimal selection method is used to select the Pareto solution sets. The results demonstrate that the optimization platform can effectively reduce the number of design variables (from 11 to 6) and improve stability, lateral comfort, vertical comfort and road friendliness, (2.78%, 1.19% and 14.71%, 3.09%, respectively). Also, a multi-parameter and multi-objective optimization platform of the VTT has certain applicability to straight-line conditions and different curve radius conditions. This research provides a feasible solution for the comprehensive performance improvement of the VTT passing through a minor curve. In addition, it also provides a reference for the optimization analysis of multi-marshaling vehicles considering the interaction between vehicles.