摘要
针对位置指纹方法室内定位时,因动态环境、多径效应和接入点(AP)不稳定导致精度低的问题,该文提出一种融合标准偏差(SD)与卡尔曼滤波(KF)的改进极限梯度提升(XGBoost)的室内定位方法。离线阶段,首先计算定位区域内AP点接收信号强度(RSS)数据的SD,然后进行降序排列筛选有效、稳定的AP集合;其次,利用卡尔曼滤波进行预处理抑制时变噪声的影响,并以此重构指纹数据库。最后,由于XGBoost算法性能易受初始参数的影响,采用粒子群优化(PSO)算法寻优得到最优的参数,并构建PSO-XGBoost定位模型。在线阶段,将目标点的数据送入已有的定位模型中预测目标的位置。实验结果表明,与其它算法相比,该算法具有良好的定位效果,在1 m、2 m和3 m定位范围内,定位精度分别提升9.1%、14.2%和18.75%。
To address the issue of low accuracy in indoor positioning using the location fingerprint method due to dynamic environments,multipath effects,and unstable access points(APs),this paper proposes an improved indoor positioning approach that integrates standard deviation(SD)and Kalman filtering(KF)with extreme gradient boosting(XGBoost).During the offline phase,the SD of the Received Signal Strength(RSS)data from APs within the positioning area is first calculated,followed by a descending order sorting to filter out effective and stable AP sets.Subsequently,the Kalman Filter is employed for preprocessing to mitigate the impact of time-varying noise,thereby reconstructing the fingerprint database.Finally,considering that the performance of the XGBoost algorithm is susceptible to initial parameters,the particle swarm optimization(PSO)algorithm is utilized to search for optimal parameters,constructing the PSO-XGBoost positioning model.In the online phase,the data of the target point is fed into the established positioning model to predict the target's location.Experimental results demonstrate that,compared to other algorithms,the proposed algorithm exhibits superior positioning performance,with positioning accuracy improvements of 9.1%,14.2%,and 18.75%within the positioning range of 1 m,2 m,and 3 m,respectively.
作者
凌凤智
刘高辉
LING Fengzhi;LIU Gaohui(School of Automation and Information Engineering,Xi'an University of Technology,Xi'an 710048,China)
出处
《导航定位学报》
北大核心
2025年第6期146-154,共9页
Journal of Navigation and Positioning
