A Two-Layer Multi-Objective Framework for Continuous Multi-UAV Patrol Scheduling Under Weather and Resource Constraints

This study addresses continuous UAV patrol scheduling for forest fire prevention in the Greater Khingan Mountains under multiple airports, UAV types, and monitoring zones. A multi-objective scheduling model is established considering UAV availability, weather conditions, runway capacity, fuel safety, ground turnaround, coordinated entry, and overlap coverage. The objectives are to minimize UAV workload-rate variance and global total fuel consumption. A two-layer solution method combining NSGA-II with temporal simulation is developed, in which the upper layer optimizes airport assignments and the lower layer performs UAV matching, takeoff- and landing-completion time calculation, UAV reuse, and constraint checking. Comparative experiments are conducted across 27 scenarios combining nine weather conditions and three airport-resource configurations, with fixed three-shift rotation and simulated annealing (SA) used as comparison methods. Each of the 1647 scheduling runs completes 52 monitoring rounds and 260 UAV sorties without reported constraint violations. NSGA-II achieves lower mean fuel consumption than SA in all 27 scenarios, while obtaining lower mean workload-rate variance in 17 scenarios. The proposed method also provides multiple Pareto solutions for different preferences between workload balance and fuel consumption.