A trust-aware explainable AI framework for cyber-resilient smart transportation infrastructure systems

Artificial Intelligence (AI) is transforming transportation infrastructure through data-driven decision-making for traffic management and pavement performance evaluation. However, integrating machine learning (ML) with Internet of Things (IoT)-enabled systems introduces challenges related to trust, robustness, and cybersecurity in smart transportation environments. This study proposes a trust-aware explainable AI framework for cyber-resilient transportation infrastructure systems. The framework employs Random Forest (RF) models for traffic flow estimation and resilient modulus prediction, while SHapley Additive exPlanations (SHAP) are used to enhance model interpretability and transparency.To ensure data reliability, anomaly detection and secure validation mechanisms are incorporated for IoT-based data streams. A resilience-driven learning approach, including adaptive retraining and fault-tolerant model design, is introduced to maintain performance under noisy and adversarial conditions. The framework is evaluated using representative datasets, and results demonstrate improved prediction accuracy, robustness, and interpretability compared to conventional approaches. The proposed method provides a scalable solution for developing secure and trustworthy AI-driven transportation infrastructure, supporting smart and sustainable construction practices.