Reliability-oriented framework for UAV-based inspection missions in modern power and energy systems.

Ensuring mission reliability is vital for the autonomous deployment of unmanned aerial vehicles (UAVs) in modern power and energy systems, particularly under spatial and operational constraints. This study presents a data-driven classification method that assesses the reliability of UAV-based inspection missions by identifying whether individual mission locations are suitable, at risk, or infeasible based on spatial and operational parameters. Leveraging the Cumulative UAV Routing Problem (CUAVRP) benchmark, four representative mission scenarios were analyzed, each characterized by unique UAV fleet sizes, sensor ranges, and endurance limits. Synthetic stress nodes were introduced to emulate edge-case conditions encountered in infrastructure inspection tasks. Each node was classified based on three categorical targets: Mission Feasibility, Coverage Reliability, and Deployment Suitability. A gradient boosting classification model was trained on spatial and operational features to determine node status. Evaluation across all scenarios yielded consistently high performance, with the cuavrp_d9_k6_r800 scenario achieving 97.05% accuracy, 96.33% precision, 97.72% recall, and 97.02% F1-score. Furthermore, incorporating physical-layer degradation factors such as signal attenuation, multipath fading, and interference is expected to enhance the realism of future reliability assessments and improve classification robustness. The proposed classification framework supports intelligent mission planning, enhances operational resilience, and facilitates automated UAV deployment strategies in critical inspection environments within the power and energy sector.