基于一种改进的神经网络的RFID室内定位算法  

RFID Indoor Localization Algorithm Based on an Improved Neural Network

作　　者：王用鑫[1,2] 

机构地区：[1]重庆大学,重庆400044 [2]重庆电子工程职业学院,重庆400044

出　　处：《科技通报》2016年第11期179-183,共5页Bulletin of Science and Technology

基　　金：重庆市教委科学技术研究项目(KJ1503010)

摘　　要：室内定位算法精度一直都是研究的重点,本文提出了一种基于粒子群算法,减聚类算法和Kmeans算法进行结合。本文首先构建室内定位RFID模型,构建定位方程,然后采用减聚类算法来避免人为干扰,通过K-means算法来形成初始化粒子群算法,最后采用粒子群算法训练RBF神经网络的所有参数,从而得到优化的输出模型,从而确定了定位最优点。仿真实验表明本文的算法可以有效的提高定位精度,降低能量消耗,提高定位精度10%。Indoor localization algorithm has always been the focus of researches, thus this paper proposes a combination of subtractive clustering algorithm and algorithm tbased on particle swarm optimization.This paper first establishes the indoor localization RFID model and localization equation, and then adopts the subtractive clustering algorithm to avoid human interference as well as forms particle swarm optimization through the algorithm and initializes the particle swarm optimization. Finally, this paper adopts the particle swarm optimization to train all the parameters of the RBF neutral network so as to get the improved output model and determine the optimal localization point. Simulation experiment shows that algorithm in this paper can effectively improve the localization accuracy by 10% and reduce energy consumption.

关 键 词：RFID 室内定位 减聚类算法 K-MEANS RBF 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]