A Validated Data-driven Subject and Vehicle Specific Nonlinear Biodynamic Model for Predicting Upper-Body Response in Vehicle Ride

Abstract

A six-degree-of-freedom (6-DOF) nonlinear lumped-parameter biodynamic model of the seated human upper body is formulated to predict human response under different unknown loading conditions. The model consists of six anatomically partitioned cascade body segments: pelvis, lower torso, central torso, upper torso, neck, and head. The joints are connected through nonlinear viscoelastic joints incorporating strain-stiffening restoring forces consistent with the nonlinear constitutive behavior of biological soft tissue. Fifteen subject-specific mechanical parameters are considered as design variables. The design variables are six joint natural frequencies, six viscous damping ratios, one Rayleigh coefficient, one nonlinear stiffening scale, and one mass scale. The subject-specific design variables are identified from three distinct vehicular loading cases using a hybrid two-phase optimization strategy. Particle Swarm Optimization (PSO) globally searches on a frequency-domain surrogate and seeds a bounded Nelder-Mead refinement for the full nonlinear Ordinary Differential Equation (ODE) response. A rigorous parameter idealization methodology, based on error-weighted geometric mean in log space, was implemented to fuse the multi-case optima into a single subject-representative consensus parameter set. The idealized model retains physical interpretability across loading scenarios and is subsequently used in forward dynamics to predict and validate the subject's experimental response. The framework constitutes a complete subject-specific biodynamic pipeline, from experimental data through optimization, idealization, and forward prediction, suitable for occupational health assessment and protective equipment design across diverse loading scenarios.