This paper proposes a multi-strategy enhanced Mirage Search Optimization algorithm integrated with Simulated Annealing, termed MSO-SA, to address the limitations of existing metaheuristics in 3D UAV path planning – namely slow convergence, susceptibility to local optima and poor adaptability across scenarios.
The algorithm first incorporates a Simulated Annealing mechanism to enhance the ability to escape local optima. An L-SHADE-derived mutation-crossover operator is then introduced to balance global exploration and local refinement, accelerating convergence. Additionally, dynamic parameter adjustment, inter-subpopulation information exchange, and an intelligent restart mechanism collectively improve population diversity, global optimization efficiency and robustness.
The performance of MSO-SA was evaluated against eight mainstream algorithms on the CEC2017 benchmark suite, where it ranked first in 105 out of 116 comparative instances, demonstrating outstanding exploratory capability. Simulation experiments in various low-altitude urban UAV path planning scenarios further show that MSO-SA generates shorter, safer, and smoother trajectories. In a 3D map containing DEM terrain, MSO-SA achieved reductions of 3.90% in mean cost fitness and 58.55% in standard deviation compared to the baseline MSO, confirming its superior convergence performance and solution stability.
Confirming its superior convergence performance and solution stability.
