多普勒雷达下的机动多目标跟踪算法

Maneuvering Multi-Target Tracking Algorithm Based on Doppler Radar

针对多普勒雷达在跟踪机动多目标过程中由于多普勒盲区(DBZ)造成量测丢失、对跟踪器性能产生严重影响这一问题,提出将最小可检测速度(MDV)带入到交互多模型广义标签多伯努利(IMM-GLMB)滤波器中,利用MDV信息抑制DBZ对跟踪器的影响。首先,通过采用基于马尔科夫分支合并策略的交互多模型(IMM)算法,解决单一运动模型的情况下无法跟踪机动目标的问题;其次,将并入MDV信息的检测概率模型带入IMM-GLMB滤波器的更新方程中,并给出了详细实现过程,利用MDV和多普勒信息来改善跟踪器性能;最后,面对目前算法需要固定航迹起始位置才可以进行跟踪的问题,提出了一种适用于广义标签多伯努利(GLMB)滤波器的自适应航迹起始算法。仿真结果表明,所提出的滤波算法在不同宽度的DBZ下都具有更好的性能表现,尤其在DBZ较小时,对滤波器的性能基本没有影响,并且所提算法在单步运行时间上有34%的提升。

To solve the problem that measurement loss which happens during tracking multiple maneuvering targets by Doppler radar due to Doppler blind zone(DBZ)can jeopardize the performance of the tracker,the minimum detectable velocity(MDV)was introduced into the interactive multi-model generalized labeled multi-Bernoulli(IMM-GLMB)filter and the MDV information was used to suppress the impact of the DBZ on the tracker.Firstly,the interactive multiple model(IMM)algorithm based on the Markov branch merging strategy was adopted to find a solution to the difficulty that the maneuvering target could not be tracked under the condition of a single motion model.Secondly,the detection probability model incorporating the MDV information was introduced into the update equation of the IMM-GLMB filter with the implementation process being detailed,and the MDV and Doppler information were used to improve the performance of the tracker.Finally,an adaptive track start algorithm suitable for the generalized labeled multi-Bernoulli(GLMB)filter was proposed with a view to solving the problem that the current algorithm needs a fixed track start position to track.The simulation results show that the proposed filtering algorithm has better performance under different DBZ widths with DBZ having little impact on the performance of the filter especially when it is small and also sees a 34%improvement in single step running time.