多传感器数据融合的数学方法研究  

Mathematical Methods of Multi-Sensor Data Fusion

作　　者：曾黎[1] 蒋沅[2] 

机构地区：[1]云南红河学院数学学院,云南蒙自661100 [2]山东大学控制科学与工程学院,山东济南250061

出　　处：《云南民族大学学报（自然科学版）》2010年第5期321-324,共4页Journal of Yunnan Minzu University:Natural Sciences Edition

基　　金：国家自然科学基金(10971120);云南省教育厅科学研究基金(07Y10100)

摘　　要：为了有效融合多个传感器的测量数据,得到准确的融合结果,综合对比了基于关系矩阵应用综合支持度的数据融合方法、基于Bayes理论的数据融合方法和基于分批估计理论的分组融合方法.提出了分批数的大小和与其他测量数据偏差比较大的数据的分配方式决定了分组融合方法融合结果的准确性,详细分析了不同分批数对融合结果的具体影响,提出了有效的数据分配方法.实例计算结果表明,合适的分批数以及对偏差比较大数据的合理分配可以有效地提高融合结果的准确性,对提高测量系统的测量准确性有很好的促进意义.For getting accurate fused data by fusing multi - sensor measurement data, this research has made a comparison among three data fusion methods： the fusion method based on the relation matrix and comprehensive support, the fusion method based on Bayes＇ theorem and the grouping fusion method based on the batch estimation theory. Through a detailed analysis of the influence on fused data caused by different batch numbers, the research has reached the conclusion that both batch numbers of the measurement data and the distribution modes of the data that deviate much from the others determine the accuracy of fused data in the grouping fusion method, and an effective distribution mode is proposed. The experimental results show that proper batch numbers and rational distribution modes can effectively improve the accuracy of fused data, which plays an important role in promoting the measurement accuracy of the industrial measurement system.

关 键 词：数据融合 关系矩阵 Bayes理论 分批估计理论 分组融合 

分 类 号：TN911.2[电子电信—通信与信息系统]