Constitutive optimization modeling of magnetorheological dampers under multiple influencing factors

Accurate constitutive modeling of magnetorheological dampers (MRDs) under multi-factor coupling remains challenging due to the inherent trade-off between experimental cost and parameter identification accuracy under multiple influencing factors. To address these limitations, this study proposes a pioneering Hippopotamus Optimization (HO) with Nelder–Mead Simplex (NMS)-Stacked Transformer (HNMS-ST) framework, which innovatively integrates three synergistic components to enhance the precision in characterizing constitutive parameters of the proposed Biplastic–Bingham (BB) model with local loss: (1) a Transformer-based data enhancement model that replenishes synthetic training data, effectively reducing experimental costs; (2) a constrained HO combined with NMS strategy and outlier penalty mechanism, which resolves parameter identification instability in the BB model under multi-factor coupling; and (3) a dynamic parameter mapping model using Transformer architecture to correlate optimized constitutive parameters with key influencing variables, including magnetic induction derived from finite element analysis (FEA) and temperature obtained by measurement. Extensive experimental verifications underscore the superior predictive accuracy and adaptability of the HNMS-ST BB model under multiple influencing factors compared to traditional methods. The proposed method resolves the parameter characterization challenges of piecewise constitutive models, thereby enabling high-precision modeling of MRD while reducing experimental costs.