Scholarlink

<p>Anthropomorphic Test Devices (ATDs) serve as surrogates for human occupants in vertical high-velocity impact to evaluate biomechanical responses. To guide the optimization of population-specific ATDs, this study proposes a nonlinear lumped-parameter dynamic model parameterized by hysteresis trajectories. The model parameters are identified by accounting for coupling among connection points along the load-transfer chain and the effects of loading rate. An ordinary differential equation (ODE)-based neural network whose topology is consistent with the load-transfer chain is developed, in which drop height is embedded as a conditioning variable to characterize loading-rate-induced dynamic variations. Drop experiments were conducted using a height-controlled rig, with loading rates varied by drop height. The measured responses were used to validate the identified model, which showed high agreement and stable during the main impact phase. Compared with EKF-based estimation and optimization-based calibration, the proposed method achieved better response reconstruction accuracy and output stability across all noise levels, while EKF performed better in parameter stability under low-noise conditions. Furthermore, to reduce the influence of population-specific segment mass distributions, the identified inter-segment connection parameters were reformulated using a local reduced-mass approach to obtain reduced-mass-normalized hysteresis characteristics, providing quantitative guidance for optimizing an in-house Chinese-body-size ATD with the Hybrid III ATD as the reference baseline.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hysteresis-trajectory parameter identification of nonlinear vertical impact dynamics for a lumped-parameter occupant model

  • 期刊论文

  • 发表日期：

    2026年7月17日

查看全文
  • Xinge Si,
  • Changan Di,
  • Peng Peng,
  • Yongjian Zhang,
  • Cong Xu

摘要

Anthropomorphic Test Devices (ATDs) serve as surrogates for human occupants in vertical high-velocity impact to evaluate biomechanical responses. To guide the optimization of population-specific ATDs, this study proposes a nonlinear lumped-parameter dynamic model parameterized by hysteresis trajectories. The model parameters are identified by accounting for coupling among connection points along the load-transfer chain and the effects of loading rate. An ordinary differential equation (ODE)-based neural network whose topology is consistent with the load-transfer chain is developed, in which drop height is embedded as a conditioning variable to characterize loading-rate-induced dynamic variations. Drop experiments were conducted using a height-controlled rig, with loading rates varied by drop height. The measured responses were used to validate the identified model, which showed high agreement and stable during the main impact phase. Compared with EKF-based estimation and optimization-based calibration, the proposed method achieved better response reconstruction accuracy and output stability across all noise levels, while EKF performed better in parameter stability under low-noise conditions. Furthermore, to reduce the influence of population-specific segment mass distributions, the identified inter-segment connection parameters were reformulated using a local reduced-mass approach to obtain reduced-mass-normalized hysteresis characteristics, providing quantitative guidance for optimizing an in-house Chinese-body-size ATD with the Hybrid III ATD as the reference baseline.