一种基于不确定度的数据关联敏感指标

A Sensitive Metric of Data Association Based on Uncertainty

数据关联是信息融合研究领域中最为关键的步骤之一,关联性能的好坏直接影响最终的融合效果。分析并指出了目前常用评估指标的不足之处,同时提出了一种基于不确定度的数据关联敏感指标。该指标将状态向量及其误差协方差视为一种概率分布,在位置均方根误差基础上,引入不确定度的概念,完成了对状态估计无偏性和稳定性的综合评价。实验仿真结果表明:针对不同的数据关联算法,由于考虑了估计状态的不确定性,该指标可以更加敏感地反映出各个关联算法之间的优劣。

Data association is one of the most important parts in information fusion, whose performance directly affects the final fusion result. By analyzing the shortcoming of these commonly used assessment metrics, this paper proposes an uncertainty degree-based sensitive metric of data association. This metric takes state variable and its error covariance as a kind of probability distribution, and further introduces the concept of uncertainty degree to attain an overall assessment on unbiassedness and stability of estimate errors. The simulation results show that by considering the uncertainty of the state estimate for different data association algorithms, the present metric can sensitively reflect their pros and cons.