摘要
为消除基于虚拟节点的数据融合算法(VNB-DF)的局部误差过大现象,设计了一种误差分级(EG)的虚拟节点数据融合算法。该算法根据目标的精度要求,设定误差等级,通过多项式拟合系数来表示一定范围内的监测数据的分布状态,在内存中生成虚拟节点。实验表明,与簇内数据取均值和VNB-DF算法相比,该算法大大提高了数据采集精度,在多种类环境应用方面能够表现出良好的性能。
In order to eliminate the phenomenon of the error is too large at local area brought by data fusion based on virtual node algorithm(VNB-DF),this paper designed a classification error(EG) the virtual node data fusion algorithm.The algorithm accorded to the target accuracy requirement,and set the error level,through the polynomial fitting coefficients to express the distribution state of monitoring data within a range,and then generated virtual node in the memory.Experimental results show that compared with the data obtained with the cluster mean and VNB-DF algorithm,the EG algorithm greatly improves the accuracy of data collection,can show a good performance in types of environment.
出处
《计算机应用研究》
CSCD
北大核心
2012年第3期1008-1010,1018,共4页
Application Research of Computers
基金
国家“863”计划资助项目(2009AA05Z203,2007AA10Z241)
国家自然基金资助项目(2009AA05Z203)
高等学校博士点新教师基金资助项目(20100093120007)
中央高校基本科研业务费专项资助项目(JUSRP11129)