In recent years, machine learning (ML) has become increasingly popular within the field of structural engineering. It has been used to provide accurate solutions for structural health monitoring, structural analysis, and structural design to name a few impacted areas. Within the domain of ML for structural engineering, there is a growing interest in the application of physics-informed machine learning (PIML). PIML is a class of ML techniques that allow preexisting physical knowledge to be injected into an ML model and provide a model that is more grounded in established principles than a purely data-driven model. This foundation in established physical knowledge is especially of interest to structural engineers, who value the real world explainability of the models they use. This study aims to systematically synthesize the literature at the intersection of PIML and structural engineering and to identify the critical research gaps to inform future directions. A scoping literature review is conducted to understand what PIML methods are being used within the domain of structural engineering and how they are being applied. From 1,358 records found in the Web of Science database, 103 studies were included based on various inclusion criteria. These findings are used to present a classification of PIML methods utilized in structural engineering and to synthesize an application-focused categorization to facilitate communication within the domain. The articles are further reviewed by algorithm type, structural engineering application, data format, and year, allowing insights into trends and patterns in the literature. Based on the findings of the review, the current concerns and challenges of PIML application in the structural and material engineering fields are discussed and future possible research directions are proposed.
Murphy et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: