多传感器模糊融合跟踪算法

针对集中式融合结构跟踪系统,利用随机逼近算法分析了权值的最优分配原则,提出了一种基于模糊推理的多传感器融合跟踪算法。该算法采用协方差匹配技术,依据滤波新息,动态调整测量噪声方差,使融合系统的均方误差始终最小。同时利用双滤波器结构,根据系统方差,实现滤波器间的动态切换,提出了基于模糊推理的并行双Unscented卡尔曼滤波自适应跟踪算法,增强当前统计模型对弱机动目标的适应能力。针对机动和非机动飞行航路进行了算法仿真,结果表明,在时变测量噪声条件下,采用模糊融合跟踪算法前后的速度均方根误差分别为45.7m/s和36.2m/s, 18.7m/s和9.6m/s,提高了多传感器系统的稳健性和跟踪精度。

Fuzzy tracking algorithm based on multi-sensor fusion

In view of the centralized multi-sensor fusion tracking system, the optimal distribution principle of weight is analyzed by the stochastic approximation algorithm and a fuzzy tracking algorithm based on multi-sensor fusion is proposed. With covariance matching technique and innovation information of filtering, the noise variance is dynamically adjusted and the mean square error of the fusion system always keeps minimum. In addition, dynamic switchover between filters is carried out based on double-filter structure and system variance. Two adaptive Unscented Kalman filters, which are working in parallel and can be switched according to the system covariance, are presented. The algorithm overcomes lower tracking precision of the current model for weak maneuvering targets. For the time-varying measurement noises and target trajectories of maneuvering and non-maneuvering, computer simulation results indicate that the RMSEs (Root Mean Square Errors) of velocities are 45.7m/s and 36.2m/s, 18.7m/s and 9.6m/s, respectively. Both the robust ability and tracking accuracy are improved.