Condition-Driven Direct-to-Satellite Multi-UAV Inspection for Offshore Wind Farms

With the expansion of offshore wind farms, the complex maritime conditions and harsh offshore environments pose challenges to routine inspections, including limited communication connectivity and poor response timeliness. In this article, a condition-driven direct-to-satellite multi-UAV inspection scheme is proposed. Deviating from conventional geometry-based routing, a novel prioritization mechanism is developed to intrinsically couple the anomalous electrical states of offshore wind turbines (OWTs) with spatial flight geometries. The scheme further improves operational efficiency by optimizing UAV trajectories under a priority-consistent sensor wake-up mapping for minimum inspection time. However, the formulated optimization problem is inherently a nonconvex mixed-integer nonlinear program (MINLP) with an unknown number of time slots, which makes conventional methods ineffective. To overcome this challenge, the problem is reformulated as a Markov decision process (MDP), and a K-means pre-partitioned soft actor-critic (KP-SAC) algorithm is developed to efficiently search for high-quality solutions. Simulation results verify that the proposed method significantly reduces inspection completion time compared with the baseline schemes under diverse sensing-load settings and communication conditions. From an offshore wind farm operation perspective, the post-processing evaluation demonstrates that reduced inspection completion time enables earlier inspection-triggered intervention, with shorter periods of degraded operating conditions and curtailed available output.