Location-routing optimization for multifunctional unmanned aerial vehicles in post-earthquake scenarios

Abstract

Damaged transportation infrastructure and delayed, making timely emergency response difficult. To address this, we develop a multifunctional UAV location–routing optimization model with time-window and energy constraints, where UAVs perform both relief delivery and damage reconnaissance. We propose an LLM-based adaptive large neighborhood search algorithm (LLM-ALNS-Par), in which an LLM configures hyperparameters to improve search efficiency and solution quality. Numerical experiments show that LLM-guided hyperparameter settings outperform Bayesian optimization and ant colony optimization. These results demonstrate the practical value of the proposed modeling and LLM-ALNS-Par framework for post-disaster UAV emergency planning.