摘要
针对复杂电磁环境下目标跟踪需要兼顾实时性和精确性的问题,应用测量领域的数据处理方法,提出了回归分析的数据关联算法.首先,利用对滤波曲线的两点回归分析,逐一预测各观测到达时刻的置信区间,筛选出关联点;接着,以系统处理周期为分组标准,对筛选序列进行成组观测数据回归分析;最后,计算出观测融合点,并用该点更新滤波器新息.该算法不仅能将复杂的关联转变成测量点迹动态更新过程,而且优化了同步化处理步骤.仿真实验表明:回归分析算法与联合概率数据关联算法相比,在直线运动场景下,两者的均方根误差及轨迹丢失率相近,且随着目标数目的增多,前者在平均占用CPU时间上的优越性更加突出;在曲线运动场景下,两者跟踪误差相当,前者占用CPU时间仅为后者的1/6.
A data association method based on regression analysis is proposed for the combined requirements of the real time and accuracy in target tracking under complex electromagnetic envi- ronment by applying the artifice dealing with the points in the measure field. Confidence intervals of the targets at a given time are predicted by calculating the regression coefficients of the system track, and the associated observations are screened out step by step. Then, an optional time is assigned as the system period, and a group regression analysis on the distilled observation arrays in the period is performed. Finally, the fused observation points are calculated and the innovations are refreshed. The method can not only simplify the complex problem of data association to the dynamic process of refreshing, but also optimize the synchronization step. Simulation results show that the proposed method and joint probability data association （JPDA） have the similar performance on the RMSE and the probability in missing the track, but the new method has supe- riority on the average CPU time when the number of targets is increased on the linear situation; and that tracking errors of both the methods are similar,but the CPU time of the new method is only 1/6 CPU time of the JPDA on the curvilinear situation.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2011年第8期92-96,107,共6页
Journal of Xi'an Jiaotong University
基金
国防预研基金资助项目(KJZ06088)
关键词
置信区间
筛选序列
观测融合点
目标跟踪
confidence intervals
distilled arrays
fused observation points
target tracking
