基于群体智能与随机有限集的复杂低空飞鸟目标跟踪方法研究

Avian radar, which could operate 24/7 automatically without limitations on such factors as the visibility, becomes the important technical means for bird strike avoidance at airport. In this subject, based on the previous research, it is planned to solve the problem of group target tracking of flying birds with the theories of swarm intelligence and random finite set (RFS) and study the maneuvering models of multi-target and group target of flying birds in complex low-altitude airspace. The method is proposed for target tracking of flying birds and false alarm rejection in strong ground clutter environment, providing a new approach for further improvement of the detection capability of the avian radar. The main research methods include: (1) for multiple targets of independent flying birds, constructing the random finite model sets for flying birds, and studying the multi-target tracking methods for flying birds with the combination of RFS based filters and random finite model sets; (2) for group target of intensive flying birds, constructing the maneuvering group target model with swarm intelligence, studying the measurement matching methods based on group target maneuvering model of flying birds and spatial distribution features, and realizing the group target tracking method with RFS based filters; (3) extending the posterior Cramer-Rao lower bounds (PCRLB) to predict the multi-target and group target tracking performance for tracking unknown time-varying number of targets, and studying the performance evaluation methods based on PCRLB for RFS based filters.

探鸟雷达不受能见度等因素的限制，能够全天候自动运行，成为机场鸟击防范的重要技术手段。本课题拟在前期研究的基础上，将群体智能与随机有限集理论应用于复杂低空飞鸟目标跟踪问题，研究复杂低空空域中的飞鸟多目标与群目标运动模型，提出强地杂波环境中的飞鸟目标跟踪与虚警剔除方法，为进一步提高探鸟雷达的探测能力提供新思路。主要研究内容包括：（1）对于分散飞行的飞鸟多目标，将飞鸟目标运动模型集构建为随机有限模型集，研究基于随机有限集滤波器与随机有限模型集的飞鸟多目标跟踪方法；（2）对于密集飞行的飞鸟群目标，构建基于群体智能的飞鸟群目标运动模型，研究基于飞鸟群目标运动模型与空域分布特征的量测匹配方法，实现基于随机有限集滤波器的群目标跟踪方法；（3）将后验克拉美罗推广到目标数目可变条件下的多目标与群目标跟踪性能预测，研究基于后验克拉美罗的随机有限集滤波器的性能评估方法。

探鸟雷达不受能见度等因素的限制，能够全天候自动运行，成为机场鸟击防范的重要技术手段。本课题在前期研究的基础上，针对飞鸟目标机动性、随机性、群体性等运动特点，将群体智能与随机有限集理论应用于复杂低空飞鸟目标跟踪问题。本课题研究了复杂低空空域中的飞鸟多目标与群目标建模方法，提出强地杂波环境中的飞鸟多目标与群目标跟踪方法，为进一步提高探鸟雷达的探测能力提供新思路。主要研究内容包括：（1）对于分散飞行的飞鸟多目标，将飞鸟目标运动模型集构建为随机有限模型集，研究基于随机有限集滤波器与随机有限模型集的飞鸟多目标跟踪方法；（2）对于密集飞行的飞鸟群目标，构建基于群体智能的飞鸟群目标运动模型，研究基于飞鸟群目标运动模型与空域分布特征的量测提取方法，实现基于随机有限集滤波器的群目标跟踪方法；（3）采用多种指标，对目标数目可变条件下的多目标与群目标跟踪算法性能进行了评估；（4）基于低空空域雷达监视系统，采用仿真和实测数据验证了算法性能，并将研究内容拓展到鸟击风险评估、智能驱鸟决策、飞鸟与无人机目标识别等领域，为项目的进一步深入研究奠定了基础。