Digital twins (DTs) are increasingly investigated for condition monitoring (CM), with validation reported primarily in numerical, laboratory and limited pilot studies. Conventional structural health monitoring (SHM) methods offer limited flexibility and predictive capability, with delayed insights reported in many infrastructure applications. DTs are commonly defined as frameworks that update virtual replicas using continuous data streams. This review presents a structured overview of current DT methodologies and their reported applications in structural CM within structural dynamics. The review focuses on critical infrastructure, including bridges, offshore platforms and aerospace systems. It also identifies challenges in large‐scale DT implementation, including computational demands, data integration complexity and cybersecurity concerns. When multiple environmental and operational hazards interact, response‐based condition inference can become nonunique and ambiguous. Several studies report improved predictive accuracy for selected AI‐enhanced DT frameworks under controlled numerical and laboratory conditions, while long‐term field validation remains scarce. Evidence supporting sustained large‐scale or long‐term field performance remains limited. The review synthesises reported developments in AI‐enabled DTs for structural dynamics and distinguishes demonstrated performance from conceptual or proposed benefits. It highlights reported capabilities and limitations, including data dependence, limited generalisability and the absence of standardised benchmarking frameworks.
Adimass et al. (2026) studied this question.