摘要
数据融合的目的是合并或者综合多个数据集,成为一个完整的性能更好的数据集。基于广义延拓插值外推法提出建立两组或两组以上方程解集和数据序列融合优化的数学模型,称之为广义融合法,此方法能充分发挥数据和量测资源的潜力,运用创新的集成和互校算法,解决数据融合面临的难题。介绍了广义融合数学模型的建立和求解方法的主要步骤;对模型中权值的分配原则做了分析;最后应用于卫星导航数据处理领域。实验证明该方法实用性好,数据平稳性好,外推稳定度高,有望成为一种新的解决数据融合问题的普适求解方法。
The purpose of data fusion is to merge or synthesize two data sets so as to make them become a better and complete data set. Based on extended inter- polation method,generalized fusion method is a method which strives to create a mathematical model that consists of two or more sets of equation and data sets.This method is not only able to bring out its full potential of data and measurement resources,but also capable of using innovative integration and mutual algorithm to solve the data fasion problem. After simulation and real measurement data analysis,conclusion can be drawn that compared with the least square algorithm,the generalized fusion method is more suitable for the repair of the data generated in the process of data fusion.This paper also introduces the main steps of the establishment and the solution of the generalized mathematical model,analyzes the distribution principle of weight in the model and its application in the field of satellite navigation data processing. Experimental results show that the method is practical and stable,and it is expected to be a new method for solving the problem of data fusion.
出处
《天文研究与技术》
CSCD
2016年第3期318-325,共8页
Astronomical Research & Technology
基金
国家自然科学基金(61271284)
广西自然科学基金(2014GXNSFAA118393)
桂林电子科技大学研究生教育创新计划项目(GDYCSZ201483)资助
关键词
数据融合
广义融合
广义延拓外推
模型求解
权值分配
卫星导航数据处理
Data fusion
Generalized fusion
Generalized extrapolation
Model solving
Weight distribution
Satellite navigation data processing
