A Systematic Review of Sensor–AI Integration in Structural Health Monitoring of Civil Buildings

Abstract

Structural health monitoring (SHM) is a component of modern civil engineering. This review analyzes the integration of sensing technologies and artificial-intelligence-based methods for damage detection, localization, classification, prognosis, and anomaly detection in buildings and civil infrastructure. The database search covered Web of Science (WoS), Scopus, and IEEE Xplore for the period 1 January 2020–31 December 2025. The initial records were 292 in WoS, 311 in Scopus, and 338 in IEEE Xplore; after applying the AI-related search constraint, the corresponding AI-SHM corpora were 71, 79, and 139 records, respectively. The combined screening and eligibility workflow produced 31 open-access studies for detailed qualitative analysis, while the task-specific performance tables synthesize the subset of studies for which the sensor type, AI model, SHM task, validation context, and performance metrics were explicitly reported. The review, therefore, interprets reported performance by SHM task and sensor modality, rather than treating heterogeneous metrics as directly comparable across different datasets and experimental conditions. The results indicate that high values reported for accelerometer-, fiber-optic-, piezoelectric transducer-, and vision-based systems are mainly obtained under controlled, benchmark, simulated, or study-specific validation conditions. Consequently, robustness, transferability to operational structures, uncertainty quantification, sensor-network design, and integration with Physics-Informed Machine Learning and Digital Twin technologies remain central research needs.