On Current Trends in Forward Model-driven SHM

Abstract

Forward model-driven approaches to Structural Health Monitoring (SHM) are a category of methods in which validated physics-based models are used to generate data for machine learning classifiers. These approaches were developed to address the lack of available damage state data; which is often a problem in many SHM contexts due to it being impractical or infeasible to collect. Many data-driven approaches to SHM are successful when the appropriate damage state data are available, however the problem of obtaining data for various damage states of interest restricts their use in industry. With this aim, several forward model-driven techniques have been developed in recent years focusing on issues in generating validated physical-models that produce damage state predictions without the need for a damage state data. This paper presents the current state of forward model-driven techniques within the literature. In addition, several key technology areas are highlighted with a demonstration of the benefits and challenges a forward model-driven framework provides.