This study aims to address the multi-performance trade-offs in the design of gas tank mounting frameworks for natural gas vehicles. The primary goal is to achieve a balanced optimization of structural lightweighting, manufacturing cost, and dynamic performance under complex high-dimensional constraints.
A systematic integrated design methodology is proposed. First, an Adaptive Surrogate Modeling (ASM) approach based on Kriging is developed, utilizing a dynamic learning function to balance global exploration and local exploitation in a 48-variable design space. Subsequently, the NSGA-II algorithm is employed to generate a Pareto front. To facilitate robust decision-making, a Composite Weighting (CW) mechanism is introduced by merging the Hesitant Fuzzy Best-Worst Method (HFBWM) for subjective expertise with the Entropy Weighting Method (EWM) for objective data insights. Finally, an improved TOPSIS (ITOPSIS) method, enhanced by Grey Relational Analysis (GRA), is applied to rank and select the optimal compromise solution.
The implementation of the proposed framework leads to significant performance enhancements. Compared to the baseline design, the optimized mounting framework reduces material cost by 28.88% and structural mass by 22.9%, while simultaneously increasing the first-order modal frequency by 36.46%. Validation against high-fidelity Finite Element (FE) simulations confirms the reliability of the surrogate-based predictions, with errors maintained below 5%.
The novelty of this work lies in the integration of a dynamic learning-based adaptive surrogate model with a hybrid MCDM framework that accounts for decision-maker hesitation using fuzzy sets. This approach effectively bridges the gap between high-dimensional engineering optimization and robust multi-criteria decision-making in complex structural systems.
