Impact Load Identification Based on Strain Measurements Using a Hierarchical Bayesian Framework

Abstract Offshore platforms are frequently subjected to impact loads in extreme marine environments, such as wave slamming and collisions. These loads are characterized by short durations and extremely high magnitudes, threatening the structural safety of platforms. Obtaining precise load information is crucial for ensuring structural safety, optimizing design, and developing effective maintenance strategies. However, direct measurement of impact loads is often infeasible due to the stochastic nature of the loads, high costs of installation and maintenance, and environmental challenges. To address this, this study develops an indirect load identification method based on a hierarchical Bayesian model, utilizing strain signals as the input. The strain-load relationship is formulated as a matrix equation through discretizing the convolution relationship between strain responses and loads. To address uncertainties caused by measurement noise, a Bayesian model is employed. A hierarchical structure is introduced to model sparsity of impact load in both spatial and temporal domains, leveraging a scale mixture of Gaussians to replace the Laplace prior. Furthermore, a greedy algorithm is implemented to identify nonzero components, enhancing computational efficiency. Numerical simulations were conducted on a plate structure, demonstrating the proposed method’s robustness and effectiveness for impact load identification.