【作者】 傅灵娇； 彭冬亮； 石义芳；

【Author】 Lingjiao Fu;Dongliang Peng;Yifang Shi;The School of Automation Hangzhou Dianzi University;

【机构】 杭州电子科技大学自动化学院；

【摘要】 有中心有记忆分布式多传感器信息融合框架利用历史融合结果更新传感器新息,并通过事件触发进行信息通信,其融合性能显著优于其他融合架构,且通信开销大幅降低,被广泛应用于多传感器协同跟踪领域。然而,由于杂波干扰和多目标呈现,导致多传感器航迹来源模糊,局部航迹和中心航迹因共用历史量测和过程噪声引发状态估计误差强相关,最终使得有中心有记忆分布式多传感器航迹融合性能快速恶化。因此,本文提出了一种面向杂波中多目标自动跟踪的有中心有记忆分布式最优航迹融合,利用航迹概率关联和基于目标存在概率的航迹自动管理,有效解决航迹来源模糊问题,设计扩维互协矩阵以解除航迹之间的强相关性,实现分布式运动学状态的最优估计。通过蒙特卡洛仿真实验与集中式点迹融合和单传感器自动跟踪进行对比,验证了所提方法的航迹管理性能及跟踪精度均远优于单传感器目标自动跟踪,且达到了集中式点迹融合的最优融合性能,但通信开销大幅降低。更多还原

【Abstract】 Multi-sensor distributed tracks fusion framework with center and memory uses historical fusion results to update sensor innovation.Information communication is triggered through events and its fusion performance is significantly better than other fusion architectures but the communication cost is greatly reduced,so it is widely used in the field of multi-sensor collaborative tracking.However,due to clutter disturbance and multi-target presence,the origin of multi-sensor track is ambiguous.Moreover,the local tracks and central tracks are strongly correlated due to the common historical measurements and process noise,leading to errors in state estimation,which quickly deteriorates the performance of the multi-sensor distributed track fusion with central and memory.Therefore,this paper proposes an optimal distributed tracks fusion with center and memory for multi-target automatic tracking in clutter.It utilizes track probability association and track automatic management based on probability of target existence to effectively solve the problem of track origin ambiguity,and designs extended-dimension cross-covariance matrices to remove the strong correlation between tracks,achieving the optimal estimation of distributed kinematic states.Through Monte Carlo simulation experiments compared with centralized fusion and single-sensor automatic tracking,it validates that the proposed method’s track management performance and tracking accuracy are much better than single-sensor automatic tracking and achieve the optimal fusion performance of centralized fusion,but with significantly reduced communication cost.更多还原