基于改进MOPSO算法的无人机山地路径规划研究

Mountainous Area Path Planning for UAVs Using Improved MOPSO Algorithm

摘要: 针对复杂山地环境下无人机路径规划存在的多目标冲突、传统多目标粒子群优化算法易早熟、全局搜索能力弱等问题，提出一种多策略融合改进的MOPSO算法。首先构建山地环境数学模型，以路径长度最短、威胁障碍碰撞代价最小为优化目标建立多目标优化模型；其次采用Tent混沌映射实现种群初始化，提升初始种群分布均匀性，引入小生境策略优化领导因子选择，避免算法陷入局部最优，结合自适应高斯变异保持种群多样性；最后通过CEC2020测试函数与山地场景仿真实验验证算法性能。结果表明，与经典MOPSO算法相比，改进算法在IGD、GD、SP指标上分别降低16.27%、53.23%、50.57%，平均路径长度缩短121 m，收敛速度更快、解集分布更均匀、路径规划效果更优，可有效满足无人机山地环境安全高效飞行需求。

Abstract: Aiming at the problems of multi-objective conflicts in UAV path planning in complex mountainous areas and the defects of traditional Multi-Objective Particle Swarm Optimization algorithm such as prematurity and weak global search ability, an improved MOPSO algorithm based on multi-strategy fusion is proposed. Firstly, a mathematical model of mountainous environment is constructed, and a multi-objective optimization model is established with the objectives of minimizing path length and threat obstacle collision cost. Secondly, Tent chaotic mapping is used to initialize the population to improve the distribution uniformity of the initial population. A niche strategy is introduced to op-timize the selection of leader factors to avoid the algorithm falling into local optimum, and adaptive Gaussian mutation is combined to maintain population diversity. Finally, the performance of the algorithm is verified by CEC2020 test functions and mountain scene simulation experiments. The results show that compared with the classical MOPSO algorithm, the improved algorithm reduces the IGD, GD and SP indexes by 16.27%, 53.23% and 50.57% respectively, shortens the average path length by 121 m, with faster convergence speed, more uniform solution set distribution and better path planning effect, which can effectively meet the requirements of safe and efficient flight of UAVs in mountainous environments.