A Hybrid AI-Fuzzy Decision Support Framework for UAV-Based Durability Assessment of Civil Infrastructure

This study proposes a hybrid fuzzy decision support framework for the durability assessment of soil and water conservation structures using UAV-acquired imagery. To address the inherent uncertainty and variability in visual inspections, a fuzzy inference classification mechanism was developed to process degradation patterns captured from aerial images. By applying fuzzy set theory and expert-informed semantic rules, the system translates ambiguous visual observations into standardized condition indices (CI), enabling robust classification of structural deterioration. The resulting CI scores are further integrated into an Analytic Network Process (ANP) model, which prioritizes maintenance factors through weighted multi-criteria reasoning. This dual-layer architecture not only enhances semantic interpretability and rating consistency, but also supports automated, large-scale infrastructure monitoring. Case results demonstrate that the fuzzy classifier significantly reduces inter-rater variation and improves classification robustness, particularly in borderline or degraded imagery scenarios. The proposed method offers a scalable, image-driven solution for infrastructure condition analysis, and exemplifies how fuzzy logic techniques can improve pattern recognition and decision support in real-world inspection applications.