基于代价函数修正的异类传感器数据关联算法

A Data Association Algorithm Based on Modified Cost Function for Heterogeneous Sensors

针对现有异类传感器数据关联算法在构造关联代价时未充分考虑最大似然估计引入误差的问题,提出了一种修正代价函数的异类传感器数据关联算法。基于主、被动传感器各自的优势信息构造观测矩阵,得到目标的位置估计,继而根据异类传感器融合系统的观测模型得到伪量测信息,并利用一阶泰勒展开近似推出伪量测的方差信息,再用其对关联代价函数进行修正,改进后的代价函数能更加准确地描述量测之间的关系。仿真结果表明,修正后的关联代价函数能更精确地反映数据关联的可能性程度,所提算法可获得更好的关联正确率。

In calculating the cost function of the current multi-target data association algorithms for heterogeneous sensors by using maximum likelihood estimation of the target position, the estimation errors are not taken into consideration.To overcome the problem, a new data association algorithm is proposed by modifying the cost function corresponding to the association hypothesis.The observing matrix is formulated to estimate the target position based on the observed range information of the radar and the angles of the passive sensors.Then the pseudo measurements can be obtained according to the measurement model of heterogeneous sensors system.The covariance information, calculated by means of the first-order Taylor series, is used to modify the association cost function.The modified cost function can reflect the correlation between measurements more reasonably owing to the integration of estimation errors.The simulation results show that the modified cost function has better performance than the original one, implying good application prospects.