Machine Learning-Driven Predictive Modeling for FRP Strengthened Structural Elements: A Review of AI-Based Damage Detection, Fatigue Prediction, and Structural Health Monitoring

The integration of Machine Learning (ML)-driven predictive modeling has revolutionized the assessment and optimization of Fiber-Reinforced Polymer (FRP) strengthened structural elements, offering advanced methodologies for damage detection, fatigue prediction, and structural health monitoring (SHM). This review provides a comprehensive analysis of AI-based predictive modeling techniques, including deep learning (DL), convolutional neural networks (CNNs), recurrent neural networks (RNNs), support vector machines (SVMs), and ensemble learning methods, for evaluating the mechanical performance and longevity of FRP-reinforced structures. The study explores how ML algorithms process sensor-acquired data from acoustic emission (AE), digital image correlation (DIC), fiber Bragg gratings (FBGs), and structural vibration measurements to predict crack initiation, fatigue failure, and progressive degradation in composite-strengthened bridges, high-rise buildings, and aerospace structures. Additionally, this review investigates thermomechanical and aeroelastic effects on FRP-strengthened elements under dynamic loading conditions, highlighting the ability of ML-based hybrid models to enhance accuracy in multi-variable stress-strain behavior prediction. The incorporation of physics-informed neural networks (PINNs) and hybrid AI-physics models further refines damage localization and severity estimation, addressing uncertainties in material anisotropy, bond degradation, and environmental aging effects. Moreover, advances in transfer learning and federated learning (FL) enable real-time SHM in large-scale infrastructure by leveraging cloud-based and edge computing frameworks for decentralized anomaly detection and predictive maintenance.This paper also discusses the integration of digital twin (DT) technology with ML-based SHM, enabling the real-time simulation, performance prediction, and life-cycle analysis of FRP-strengthened structures. Challenges such as model interpretability, data scarcity, and computational efficiency are examined, along with the potential of explainable AI (XAI), uncertainty quantification (UQ), and reinforcement learning (RL) in optimizing decision-making processes for infrastructure sustainability. The review concludes by identifying future research directions in hybrid AI methodologies, adaptive learning frameworks, and quantum-enhanced predictive modeling, aiming to enhance the resilience and durability of FRP-strengthened structural systems in aerospace and civil engineering applications.