Longitudinal type-line optimization of high-speed train for low aerodynamic noise

The basic head shape of high-speed train is determined by its longitudinal type-line （LTL）, so it is crucial to optimize its aerodynamic performance. Based on the parametric modeling of LTL constructed by non-uniform relational B-spline （NURBS） and the fluctuation pressure obtained by large eddy simulation （LES）, the Kriging surrogate model （KSM） of LTL was constructed for low aerodynamic noise, and the accuracy of the KSM was improved gradually by adding the sample point with maximum expected improvement （EI） and the optimal point from optimization. The optimal objective was searched with genetic algorithm （GA）. The results show that the total fluctuation pressure level （FPL） of the optimal LTL can be 8.7 dB less than that of original one, and the shape optimization method is feasible for low aerodynamic noise design.