无线电工程

针对传统灰狼优化(Grey Wolf Optimization, GWO)算法求解无人机三维路径规划问题时会出现收敛速度慢、容易陷入局部最优等问题，提出一种改进混合灰狼优化算法——CLGWO。基于Cat混沌映射和反向学习策略初始化灰狼种群，为算法全局搜索过程中丰富种群多样性奠定基础；提出新型非线性收敛因子的改进策略，提高算法全局搜索能力。在灰狼位置更新中提出引入狮群优化(Lion Swarm Optimization, LSO)算法的扰动因子和动态权重，使灰狼具有主动的搜索能力，避免因灰狼失去种群多样性而陷入局部最优。为验证改进算法的有效性，进行了8个国际通用的标准测试函数收敛性对比实验和无人机三维路径规划仿真实验。实验结果表明，CLGWO算法在单峰、多峰函数上均有较好的收敛性、较高的寻优精度；三维路径仿真环境下，CLGWO算法的平均路径长度、平均迭代次数、平均运行时间相比于GWO算法分别优化了33%、31%、52%,且路径转折少，能较好地得到全局最优值，验证了CLGWO算法的有效性。

To address the problems of slow convergence and falling into local optimum easily when solving UAV 3D path planning problems by the traditional Grey Wolf Optimization(GWO) algorithm, an improved hybrid GWO—CLGWO algorithm is proposed. Firstly, the gray wolf population is initialized based on Cat chaotic mapping and backward learning strategy, which lays the foundation for enriching the population diversity in the global search process of the algorithm. An improvement strategy of a new nonlinear convergence factor is proposed to improve the global search capability of the algorithm. Secondly, the introduction of perturbation factors and dynamic weights of the Lion Swarm Optimization(LSO) algorithm is proposed in the gray wolf position update to make the gray wolf have active search ability and avoid the gray wolf losing population diversity and falling into local optimum. Finally, to verify the effectiveness of the improved algorithm, eight internationally used standard test function convergence comparison experiments and UAV 3D path planning simulation experiments are conducted. The experimental results show that the CLGWO algorithm has better convergence and higher optimization-seeking accuracy on single-peak and multi-peak functions; the average path length, average number of iterations, and average running time of the CLGWO algorithm are optimized by 33%, 31%, and 52%, respectively compared with the GWO algorithm under the 3D path simulation environment, and there are fewer path transitions, which can better obtain the global optimum. The simulation results verify the effectiveness of CLGWO algorithm.