Hybrid multi-objective robust design optimization of a truck cab considering fatigue life

Abstract

Fatigue performance optimization without considering uncertainties of design variables can be problematic or even dangerous in real life. In this paper, a hybrid multi-objective robust design optimization methodology is proposed to make a proper tradeoff between the lightweight and fatigue durability for the design of a truck cab. However, the uncertainties, in reality, could lead to the optimized design unstable or even useless; this situation can be more serious in non-deterministic optimization. The Taguchi robust parametric design technique is adopted to refine the intervals of design variables for the subsequent optimization based on the validated simulation model against fatigue tests. Three types of dual surrogate models, namely the dual polynomial response surface, dual Kriging, and dual radial basis function methods are compared, and the dual Kriging is selected to model the mean and standard deviation of the mass and fatigue life for its high accuracy. The multi-objective particle swarm optimization algorithm is utilized to perform robust design. The Pareto fronts with different weight factors are analyzed to provide some insightful information on optimum designs. The robust optimization results demonstrate that the optimized design improves the fatigue life and reduces the mass of the truck cab significantly and becomes less sensitive to uncertainty. Different optimums can be obtained based on three different normalization techniques (Linear, vector, and LMM) and three MCDM methods (TOPSIS, WPM, and WSM) from the same Pareto front. The comparison analysis emphasizes the importance of normalization and MCDM method selection in the optimal design selection process.