论文部分内容阅读
在多传感器融合中,航迹与航迹融合占有重要的地位。人们在这方面做了大量工作,丛氏等人[5-7]给出了任意通信模式下的最优融合公式。对于确定性来说,该公式是最优的,这里指的确定性是:过程噪声为零或使用全速率通信(即两传感器每接到一次新数据就通信一次)。但在实际操作中,因目标机动而不能完全忽略过程噪声;或者为节约通信宽带,传感器间不采用全速率通信。这两种情况下,系统都存在公共过程噪声,因此两传感器的量测不是条件(给定目标预先状态)独立的,所得融合公式[7]只是近似最优。文献[1]中也谈到这种情况,作者推导出了一个公式来计算不同传感器的两条航迹估计的协方差、基于[1]的结果,文献[2]考虑了两个传感器航迹估计的相关性,并得到一个融合公式来组合局部估计。遗憾的是,文献[2]中进行贝叶斯推导时,所做的假设并不符合实际。本文中,我们指明[2]中结果潜在的近似性,并证明该结果只在ML{最大似然}意义下最优。然后,我们提出一种性能评估方法来研究各种航迹与航迹融合方法的性能。其结果给出各种操作条件下不同融合方法的性能范围。
In the multi-sensor fusion, the track and track fusion occupy an important position. People have done a lot of work in this area. Cong [5] and [7] have given the optimal fusion formula in arbitrary communication mode. The formula is optimal for certainty. The certainty here is that the process noise is zero or communicates at full rate (ie, the two sensors communicate once with each new data). However, in practice, process noise can not be completely ignored due to the target maneuver, or full-rate communication between sensors is not used to save communication bandwidth. In both cases, there is a common process noise in the system, so the measurement of both sensors is not conditional (given the target’s pre-state) independently, and the resulting fusion equation [7] is only approximately optimal. In [1], we also talk about this situation. The author deduced a formula to calculate the covariance of two trajectory estimates of different sensors. Based on the results of [1], [2] considered two sensor tracks Estimate the correlation and get a fusion formula to combine the local estimates. Unfortunately, the assumption made in Bayesian derivation in [2] is not realistic. In this paper, we show the potential approximation of the results in [2] and prove that the result is only optimal in the sense of ML {maximum likelihood. Then, we propose a performance evaluation method to study the performance of various flight path and track fusion methods. The results give the range of performance for different fusion methods under various operating conditions.