摘要
针对大型目标区域室内定位中定位精度低及信号间干扰等问题,提出一种融合概率分类模型的精细化定位方法。对于大型定位区域,采用以RP的信号强度对目标区域聚类分区,缩小信号变化区域。信号的多径效应带来信号依据下的错误分区,该文采用汤普森检验理论对分区参考点进行物理坐标校验以剔除异常点。为减小AP间信号干扰,引入多元高斯的朴素贝叶斯模型对不同AP信号建立高斯概率模型,融合一对多支持向量机概率模型以提高算法定位精度。测试分析表明,该算法实现较好的聚类分区,降低信号间干扰,平均定位精度达到0.4008m,相比于其他传统算法,定位精度提高27%以上。
For the low positioning accuracy and mutual interference between signals in large target area,this paper proposes a refined positioning method based on fusion probability classification model.In order to reduce the target area of the point to be measured and narrow the range of signal changes,the system uses the received signal strength of reference point to construct a cluster partition model.The multipath effect of the signal brings wrong division under the signal basis,the system uses Thompsom test theory to check the physical coordinates of the reference points in partition,which can eliminate abnormal points.The proposed method establish a conditional probability function based on multivariate Gaussian naive Bayesian model,which can reduce the mutual interference between the signal strength of access point.To improve the positioning accuracy of the algorithm,this paper merge one-vs-rest support vector machine with multivariate Gaussian mixture model.Experimental result shows that the proposed method can achieve better signal clustering and reduce the mutual interference between access point.The average positioning accuracy reaches 0.4008m,which is more than 27%higher than other traditional algorithms.
作者
孙顺远
朱红洲
SUN Shunyuan;ZHU Hongzhou(Key Laboratory of Advanced Process Control for Light Industry(Ministry of Education),Jiangnan University,Wuxi 214122;School of Internet of Things Engineering,Jiangnan University,Wuxi 214122)
出处
《计算机与数字工程》
2023年第5期1059-1064,共6页
Computer & Digital Engineering
关键词
聚类分区
异常点剔除
多元高斯贝叶斯模型
一对多支持向量机
线性加权
cluster partition
outlier elimination
multivariate gaussian bayes model
one-vs-rest support vector machine
linear weighting
