基于LMBP神经网络的无线室内定位方法  

Wireless indoor positioning method based on LMBP neural network

作　　者：邹自明[1] 何文斌 杨小平 ZOU Zi-ming;HE Wen-bin;YANG Xiao-ping(College of Mechanical and Control Engineering,Guilin University of Technology,Guilin 51004,China)

机构地区：[1]桂林理工大学机械与控制工程学院,广西桂林541004

出　　处：《桂林理工大学学报》2018年第3期574-578,共5页Journal of Guilin University of Technology

基　　金：国家高技术研究发展计划(863计划)项目(2013AA122105)

摘　　要：无线信号在复杂环境多变的室内传播时,信号传播往往发生反射、多径和障碍物阻挡等,导致信号传播模型参数的变化。由于传统的RSSI的定位误差较大,标准BP神经网络实现室内定位训练样本时存在目标误差下降速度较慢的问题,本文采用了LMBP算法,使用Matlab软件编写LMBP神经网络算法代码进行性能仿真分析。实验结果显示,LMBP神经网络有效地解决了训练误差收敛速度慢的不足,提高了定位精度。Generally wireless signals are always diffracted and refracted while transmitting in a complex and transformable indoor circumstance,which causes the change of parameters of signal transmitting model.To traditional RSSI in positioning error,some researchers applied back-propagation artificial neural network to solve the indoor location.To overcover the shortages of the algorithm when target errors of the standard BP neural network fall too slow during processing indoor positioning of training sample,LMBP will be applied in this test.Matlab software will be used to write a LMBP neural network algorithm code to proceed simulation analysis of its performance.The result of the experiment shows that LMBP neural network effectively solves the slow convergence speed of training errors and improves location accuracy.

关 键 词：室内定位 RSSI LMBP神经网络 

分 类 号：TP520.6020[自动化与计算机技术]