摘要
本文研究了一类具有不同采样率的分布式多传感器动态系统的数据融合问题,针对一类采样率呈有理数倍关系的动态系统,提出一种基于多源异步采样数据的新融合算法.新算法首先是将来自各个传感器的测量值在融合中心的坐标系中和时钟下进行映射统一;其次,以对目标状态下一时刻的预测值与目标在该时刻状态的估计值之差为基础,建立起描述该融合周期内各个观测点处的目标状态向量之间的动态模型;然后,以该时刻目标状态基于全局信息的估计值为条件,结合建立的新模型和传统的K a lm an滤波器,利用本周期内按序到达的各传感器观测值,依次对各个观测点处目标的状态进行估计和更新;最后,在顺序得到本周期内各个观测点处目标估计值的同时,也将获得下一时刻目标状态基于全局信息的估计值或预测估计值.文中在给出新算法基本思想的同时,也较为详细地对融合算法进行了推导,并通过计算机仿真的方法,将新算法与基于时间校准的算法在估计精确度上进行了比较,从而验证了新算法的有效性.
This paper explores data fusion of distributed multisensor dynamic systems, these sensors hold different sampling rates. For the proportion between them is usually rational;a new fusion algorithm based on asynchronous sampling data is proposed. Firstly＇, the new algorithm maps and unifies all measurements in the reference frame and clock with fusion centre. Secondly,using the difference between predict value to object state of next time and state esti- mate value of this time, we establish the dynamic model between object state vector of every sampling point in the fusion period. Thirdly, combining the new established model with traditional Kalman filter, every state in this period can be estimated and updated by obtaining orderly measures. Finally, the next state estimate or predicted estimate may be got by global information after all state estimates relative to all observation point in this period have been obtained in turn. With introducing the basic idea of the new algorithm, the processes to win it are presented step by step. Using of computer simulation in terms of comparing the results utilizing the new algorithm with those based on time calibrated method via ＂estimate accuracy, the good performance arising from this new approach has been effectively validated.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2006年第3期543-548,共6页
Acta Electronica Sinica
基金
国家自然科学基金(No.60434020
No.60572051)
教育部科学技术研究项目(No.205092)
河南省国际合作项目(No.0446650006)
