Time–Frequency Identification of Peridynamic Parameters for Structural Health Monitoring Under Extreme Excitation

Abstract

Structural monitoring is an active field of research that aims to provide knowledge to preserve the built environment. Commonly, structural health monitoring involves mechanical characterization and damage identification steps that require the assumption of physical laws describing structural behaviour. Data-driven and model-driven approaches have been pursued for this purpose. However, while the former present problems in allowing an in-depth understanding of the physics behind the structural behaviour, the latter are characterized by problems related to uncertainty and the discrepancy of the model compared to reality. Peridynamics is a modern nonlocal theory capable of managing long-range forces and discontinuities in materials, which generalizes the classical continuum and particle theory in a single theory. In this paper, Peridynamics is used to bridge the gap between data- and model-driven approaches and define synergistic approaches to structural health monitoring since the theory is able to describe a wider range of complex nonlinear behaviours related to brittle collapse, such as rocking and cracking, with a smaller number of physical parameters. In the paper, the author proposes a new method for damage identification through the time–frequency distribution estimate of peridynamic parameters, where parameter values are also provided over the frequency domain, expanding the time-domain estimate. The proposed framework is validated through single- and multi-degree-of-freedom experimental benchmarks subject to abrupt stiffness degradation. While the current validation is intentionally limited to a controlled system, the methodology is formulated for generic peridynamic discretizations and is therefore potentially extendable to more complex structural configurations.