摘要
针对观测统计量的联合概率分布未知的多传感器分布式估计融合系统,作者利用Megalooikonomou等提出的直和估计思想,基于C-均值聚类方法设计了一种量化器,改进了他们基于回归树设计量化器的融合效果.进一步,在系统整体优化条件下,作者同时考虑了各分站量化器的设计,建立了估计融合模型并通过实验验证了该模型的有效性.
In this paper, utilizing the advantages of the direct sum estimation fusion method presented by Megalooikonomou et al, the authors design a new quantization method based on the C-means clustering for the multi-sensor distributed estimation fusion system without any statistical information other than the observations of all sensors. The proposed method has remarkably improved the fusion performance comparison to the method based on the regression trees. Further, a new estimation fusion model is built in which the quantizers of all sensors are considered simultaneously on condition that the system is optimal. The experiments show the effectiveness of the presented model.
出处
《四川大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第5期1241-1244,共4页
Journal of Sichuan University(Natural Science Edition)
基金
国家自然科学基金(60574032)
四川省应用基础研究项目(05JY029-019-2)
关键词
分布式估计
C-均值聚类
直和估计
distributed estimation, C-means clustering, direct sum estimation
