估计融合算法的一类敏感指标研究

A Sensitive Metrics of Estimation Fusion Algorithms

为了改善在复杂多变的环境下多源信息融合的准确性和鲁棒性,可在融合过程中进行算法管理,自适应配置算法的方法实现。在算法配置中反馈所需的条件采用算法敏感指标进行评价。在状态估计敏感指标OSPA距离测度基础上,提出估计融合中的敏感指标GOSPA距离测度,GOSPA距离将真实航迹和全局航迹之间的误差分离成航迹距离误差和航迹关联误差。通过对比实验表明,估计融合中的全局OSPA测度指标对航迹融合算法的性能评估是敏感的。

The adaptive algorithm configuration in algorithm management can improve the precision and robustness of multi-source information fusion in complex and changing environments.The configuration depends on some evaluating metrics which is sensitive on different algorithms.Based on the sensitive metrics OSPA of state estimation algorithms,a sensitive metrics GOSPA of estimation fusion algorithms is proposed,which computes the deviation between the truth and the global track.GOSPA can adjust the weights on deviations of track distance and track association.The experiments show that GOSPA is sensitive on performance evaluation of different estimation fusion algorithms.