ROBUST DESIGN OPTIMIZATION OF THE DYNAMIC RESPONSES OF A TRACKED VEHICLE SYSTEM

This paper presents the robust design optimization of the dynamic responses of a heavy military tracked vehicle system. The tracked vehicle model addressed in this study has 954 degrees of freedom and consists of 189 bodies in total: 37 bodies for the chassis, such as sprockets, road wheels, road arms, etc.; 76 track link bodies for each track subsystem; 36 revolute joints; and 152 bushing elements. The design objectives were to minimize the maximum vertical acceleration of the hull and its variance while satisfying the wheel travel constraints for torsion bars and the hydro-pneumatic suspension units within +/- 1 sigma ranges. To avoid the difficulty of the design sensitivity analysis and to overcome the numerical noise, a progressive meta-model technique was employed in the optimization process. First, space-filling methods were used to determine the minimum number of sample points. Second, the simultaneous kriging method was used to construct the initial meta-models, and the augmented Lagrange multiplier (ALM) method was then used to solve the robust design problems of the meta-models. Third, the new design results were added to the analysis results for the initial sample points, and the meta-models were updated automatically. Next, the optimizer resolved the robust design problems of the updated meta-models. These processes were repeated until the convergence tolerances were satisfied. The robust design optimization of the tracked vehicle system, with 11 random design variables, was solved in only 26 analyses, including 12 analyses for the initial meta-models and 14 analyses added during the iterative optimization process.