Ship-emission monitoring in construction-stage port waters: Prescribed plume sampling and mothership-supported Multi-UAV planning

Ship-emission monitoring in construction-stage port waters requires coordination between local UAV sampling and mothership-supported operations under endurance and navigation constraints. This study develops a two-stage computational planning framework for prescribed plume-sampling tasks and multi-UAV scheduling. Offline plume simulations compare spiral-ascent and diagonal-zigzag scans, yielding simulated capture rates of 98.261% and 81.237% under the modeled criterion. The selected complete scans, with a per-vessel adjustment allowance, define heterogeneous service budgets. A wave-based model coordinates UAV task chains, recovery locations, battery replacement and mothership obstacle-avoiding travel. The proposed Wave-Adaptive Large Neighborhood Search (Wave-ALNS) directly reconstructs wave and chain structures. Across nine AIS-derived cases with 30, 40 and 50 vessels, Wave-ALNS reduces mean mission time by 1.142–5.475% relative to GA under matched solver-time budgets and by 2.521–13.491% relative to sequence ALNS with a 300 s budget. Its mean solver time is 7.699–29.733 s. Comparisons with generic operators and alternative service-time configurations further examine the contribution of scheduling structure and prescribed task workloads.