Impact load identification for stiffened ship hull plates using TSVD-GCV regularization

Abstract

Identification of transient impact loads on stiffened ship hull plates is critical for structural safety assessment, yet this is typically an ill-posed inverse problem. This paper proposes a time-domain load identification method based on the unit impulse superposition principle. A numerical-experimental hybrid strategy is employed, where impact tests on a stiffened plate model were systematically designed and conducted to acquire real strain responses, combined with the transfer matrix constructed via finite element simulations. To address the ill-posedness inherent in the inverse process, the Truncated Singular Value Decomposition (TSVD) method with the Generalized Cross-Validation (GCV) criterion is adopted. Furthermore, the influence of the calculation time step is investigated by defining a dimensionless ratio of the time step to the load rise time, and the optimal ratio balancing accuracy and stability is identified. Crucially, to mitigate the spatial correlation of structural responses, a sensor placement optimization strategy based on breaking spatial symmetry is proposed. By reducing information redundancy and the condition number of the transfer matrix, this asymmetric arrangement significantly enhances numerical stability. Experimental validation demonstrates that the proposed method achieves high accuracy in reconstructing the time histories of impact loads at different locations.