The accurate identification of substructure physical parameters is crucial for predicting system dynamic behavior, yet it poses significant challenges under limited measurement conditions. To address the issue of incomplete measurements, this paper proposes a novel VMD-SINDy method that integrates Variational Mode Decomposition (VMD) with Sparse Identification of Nonlinear Dynamics (SINDy). The proposed approach first employs VMD to transform the limited measurement signals from the physical space into a more informative modal space, thereby retaining the implicit influence of unmeasured substructures. A physically constrained function library is then constructed in modal space based on the prior model, enabling the identification of substructure physical parameters via the SINDy algorithm. The proposed method is validated on a ground experimental platform for a solar array attitude adjustment system, where the Solar Array Drive Assembly (SADA) serves as the identification target, explicitly characterized as a nonlinear substructure comprising a stepper motor and a harmonic drive. Global uncertainty analysis and multi-condition cross-validation demonstrate that the identified parameters reliably capture the substructure's physical properties despite limited measurement data. Furthermore, the on-orbit steady-state response, derived by mathematically decoupling ground suspension effects, reveals significant structural hardening behavior, providing a critical dynamic baseline for system-level behavior prediction.
Zhang et al. (Thu,) studied this question.