CAJ | 학술논문

针对多普勒盲区条件下预警机雷达多目标跟踪问题，基于交互式多模型(IMM，Interacting Multiple Models)、联合概率数据互联(JPDA，Joint Probability Data Association)和分布式不敏卡尔曼滤波(UKF，Unscented Kalman Filter)提出了预警机雷达与地基雷达对目标进行协同跟踪的方法。该方法利用目标的状态估计和预测实时计算每部雷达的动态融合权值，预测目标的多普勒频率。当预警机雷达对目标的量测不存在且检测到目标进入预警机雷达多普勒盲区时，由预警机雷达对目标状态进行外推,以此产生虚拟量测,用虚拟量测与地基雷达协同跟踪对目标的融合估计状态进行更新；若预警机雷达对目标的量测不存在且目标不是进入多普勒盲区时,由地基雷达单独对目标的融合估计状态进行更新。当目标飞出预警机雷达多普勒盲区后，将预警机雷达对目标的状态估计再次与地基雷达进行关联，并根据动态权值融合更新目标状态。仿真结果表明，该方法能够改善多普勒盲区内多目标航迹的连续性和跟踪精度。
침대다보륵맹구조건하예경궤뢰체다목표근종문제，기우교호식다모형(IMM，Interacting Multiple Models)、연합개솔수거호련(JPDA，Joint Probability Data Association)화분포식불민잡이만려파(UKF，Unscented Kalman Filter)제출료예경궤뢰체여지기뢰체대목표진행협동근종적방법。해방법이용목표적상태고계화예측실시계산매부뢰체적동태융합권치，예측목표적다보륵빈솔。당예경궤뢰체대목표적량측불존재차검측도목표진입예경궤뢰체다보륵맹구시，유예경궤뢰체대목표상태진행외추,이차산생허의량측,용허의량측여지기뢰체협동근종대목표적융합고계상태진행경신；약예경궤뢰체대목표적량측불존재차목표불시진입다보륵맹구시,유지기뢰체단독대목표적융합고계상태진행경신。당목표비출예경궤뢰체다보륵맹구후，장예경궤뢰체대목표적상태고계재차여지기뢰체진행관련，병근거동태권치융합경신목표상태。방진결과표명，해방법능구개선다보륵맹구내다목표항적적련속성화근종정도。
Focusing on the tracking problem of multi-target hidden in AEW (airborne early warning) radar DBZ(Doppler blind zone), a collaborative tracking technique between AEW and ground-based radar was proposed based on IMM, distributed UKF and JPDA. The dynamic fusion weights were calculated for all radars and the Doppler frequency of each target was predicted in real time using target state estimation and prediction. On one hand, as long as the target measurements of AEW radar did not exist and the target was hidden in DBZ of AEW radar judged from its predicted Doppler frequency, targets state would be extrapolated by AEW radar, which creates virtual measurements. The target fusion state estimation would be updated by both virtual measurements of AEW radar and real measurements of ground-based radar. On the other hand, when the targets measurement of AEW radar did not exist and the target was not hidden in DBZ of AEW radar judged from its predicted Doppler frequency, the targets fusion state will be updated by ground-based radar all alone. After the target fly off the DBZ, the target state estimation of AEW radar will be associated with that of ground-based radar again. And the targets fusion state will be updated by both AEW radar and ground-based radar according to the dynamic weights. Simulation results show that the presented method can improve the continuity and accuracy of multi-target tracking in DBZ.