用于UAV运动目标搜索的自适应粒子群算法

Adaptive Particle Swarm Optimization for UAV Moving Target Search

针对运动编码粒子群算法在处理无人机运动目标搜索问题时存在被其他高概率区域混淆、算法搜索成功率不够高的问题,提出了一种基于运动编码的自适应学习策略粒子群优化算法以优化无人机飞行路径。该算法先设计了适应于各种搜索场景的初始化方案;再融入聚类算法用以动态划分粒子群,并改进了子群中不同类型粒子的更新方程以适应路径规划中的粒子子群;最后添加了自适应学习策略以控制参数,旨在保持收敛速度的基础上提高搜索到最优路径的概率。在不同搜索场景下的实验结果表明,与运动编码粒子群优化算法相比,算法的检测性能提升了6%。此外,与其他元启发式优化算法的对比结果也展示了算法的优势。

The motion encoded particle swarm algorithm has encountered difficulties,such as being confused by other high probability regions and the search success rate is not high enough in dealing with the unmanned aerial vehicle(UAV)motion target search problem.In this paper,an adaptive learning strategy particle swarm optimization algorithm based on motion encoding is proposed to optimize the UAV flight path.The algorithm first designs an initialization scheme applicable to various search scenarios.Then,a clustering algorithm is incorporated to dynamically partition the particle swarm,and the update equation of different types of particles in the sub-swarm is improved to adapt to the particle sub-swarm in path planning.Finally,an adaptive learning strategy is added to control the parameters,aiming to increase the probability of searching for the optimal path while maintaining the convergence speed.Experimental results in different search scenarios show that the detection performance of the algorithm is improved by 6% compared to the motion-coded particle swarm optimization algorithm.In addition,comparison results with other metaheuristic optimization algorithms demonstrate the advantages of the algorithm.