多无源传感器去相关数据关联算法
鹿传国;冯新喜;孔云波;张迪
【期刊名称】《自动化学报》 【年(卷),期】2014(000)003
【摘要】对基于多维分配模型的多无源传感器(Multi-passive-sensor system, MPSS)多目标数据关联算法进行了归纳分析,指出该模型不仅忽略了极大似然估计所引入的随机误差,而且未充分考虑量测与伪量测之间的相关性。继而建立了一种去相关修正数据关联模型,并提出利用无迹变换计算二者之间的互协方差。另外定义了概念“解的区分度”来评估关联代价构造的合理性。最后进行了仿真实验,结果表明去相关后的关联代价能更精准地反映数据关联的可能性,所提关联算法运算时间有所增加,但关联性能更佳。ˉter summarizing and analyzing the multi-target data association algorithms based on the S-D assignment for multi-passive-sensor system, it is pointed out that the association algorithms above have ignored both the error introduced by the maximum likelihood estimation and the relativity between the measurements and the pseudo ones. Then, a decorrelation-based data association model is built and the unscented transform is proposed to compute the mutual covariance between measurements and the pseudo ones. Meanwhile, a new concept, the discrimination of answers, is defined to evaluate the association cost forming methods. Lastly, results of simulation have shown that the uncorrelated cost function can reflect the association probability more accurately and the proposed
多无源传感器去相关数据关联算法
