小波包分解与长短期记忆融合的超宽带无线测距算法  

Ultra-wide band wireless ranging algorithm based on wavelet packet decomposition and long short-term memory fusion

作　　者：崔学荣[1] 于寒蕾 李娟[1] 姜斌 李世宝[1] 刘建航[1] CUI Xuerong;YU Hanlei;LI Juan;JIANG Bin;LI Shibao;LIU Jianhang(China University of Petroleum(East China),Qingdao,Shandong 266000,China)

机构地区：[1]中国石油大学(华东),山东青岛266000

出　　处：《导航定位学报》2023年第6期102-109,共8页Journal of Navigation and Positioning

基　　金：国家自然科学基金项目(61902431)。

摘　　要：针对超宽带(UWB)无线定位标准虽然已经取得了很大的发展,但是由于存在多径效应和非视距等影响,导致直达径分量到达时间估计存在很大误差,因而定位精度不高的问题,提出小波包分解和长短期记忆融合的超宽带无线测距算法:采用小波包分解节点能量接收滤波,对接收到的信号进行分解,用最后一层各节点的能量占总能量的比值构建能量谱特征向量;然后用长短期记忆网络建立能量谱特征向量和实际距离的映射关系,从而实现基于UWB的测距和定位,为后续定位提供基础。实验结果表明,结合长短期记忆网络和小波包能量检测的测距算法比传统的能量检测方法具有更大的优势,其测距定位精度更高,更稳定。Aiming at the problem that although ultra-wide band(UWB)wireless positioning has achieved great development,the influence of multipath effect and non line of sight(NLOS)leads to large errors in the estimation of the arrival time of the direct path component,thus the positioning accuracy is low,the paper proposed a UWB wireless ranging algorithm based on wavelet packet decomposition and long short-term memory fusion:wavelet packet decomposition node energy reception filtering was used to decompose the received signal,and the energy spectrum feature vector was constructed with the ratio of the energy of each node in the last layer to the total energy;and the mapping relationship between the feature vector of energy spectrum and the actual distance was established by using long and short term memory network,so as to realize the ranging and localization based on UWB for providing the basis for subsequent positioning.Experimental result showed that the ranging algorithm combining long and short term memory network and wavelet packet energy detection,with its higher ranging and positioning accuracy and more stable performance,could have more advantages than the traditional energy detection method.

关 键 词：室内定位 超宽带(UWB) 小波包分解 长短期记忆网络 

分 类 号：P228[天文地球—大地测量学与测量工程]