Hybrid data + model‐based submodeling method for a refined response estimation at critical locations

Abstract

The complex geometry of structural components results in uneven stress distribution in structures under loading. The regions with stress concentrations often act as critical locations from where damages can initiate and propagate under different types of loading. Accurate estimation of stress distribution, especially at such critical locations, is vital for a more reliable prediction of possible damage prognosis or remaining useful life (RUL) estimation of the structure or a component. Traditional sensing methods often provide response measurements at a few localized points and demand sensor deployment in large numbers to get a distributed response. This may yet be sparse for the steep changes in stress at critical locations. In this study, we propose a hybrid data + model-based submodeling (HDMS) method to achieve a refined estimate of distributed structural response in and around the critical locations. The HDMS method uses just the measured response on the pre-selected boundaries around the locations of interest, as input to drive the corresponding submodel of a structural component or a connection, given its geometry and material properties are known. The performance of the HDMS method in response estimation is demonstrated through two vertically loaded plates, one with two holes connected by a slit and the other with two wide slits, respectively. The refined response estimated by HDMS could determine asymmetric response, nonlinear behavior, and permanent set at the critical locations with an average error less than 50 strain at higher load stages, making HDMS a versatile method for refined response estimation.