基于UKF-SOFNN的近距机动目标跟踪滤波算法

Filtering and Tracking Algorithm of Near-Distance Maneuvering Targets Based on UKF-SOFNN Method

对于多目标杂波环境中的机动目标跟踪,由于目标集群中各个目标间的空间距离可能小于探测器的空间分辨率,因而可能出现误跟、诱饵欺骗与杂波虚警等一系列严重后果。对此,提出一种综合运用UKF(不敏卡尔曼滤波)和SOFNN(自组织模糊神经网络)的UKF-SOFNN滤波跟踪算法,将机动目标模型视作严格的非线性系统,利用UKF-SOFNN对非线性参数的辨识能力提高对锁定机动目标的跟踪能力。仿真实例表明,该算法能有效地辨识目标群中的目标,并进行可靠的跟踪。

Tracking multi-target in clutter environment, because the distance between each target in target set may be less than the sensors resolution of explorer, problems of bug-tracking, bait deceiving and clutter false alarm emerge. So a new filtering and tracking method named UKF-SOFNN is presented, in which the models of maneuvering targets are viewed as straight nonlinear system, and better capa the UKF-SOFNN for discriminating nonlinear parameters is insured by the models. The simulation that the new method is able to resolve the specific target in target set and track it feasibly.