文献类型：期刊文献

中文题名：和声搜索算法优化神经网络的无线网络室内定位

英文题名：Indoor positioning of wireless network based on harmony search algorithm optimizing neural network

作者：付思源[1];王华东[2]

第一作者：付思源

机构：[1]甘肃中医药大学定西校区,甘肃定西743000;[2]周口师范学院计算机科学与技术学院,河南周口466001

第一机构：甘肃中医药大学定西校区

年份：2017

卷号：41

期号：4

起止页码：428

中文期刊名：南京理工大学学报

外文期刊名：Journal of Nanjing University of Science and Technology

收录：CSTPCD;;Scopus;北大核心:【北大核心2014】；CSCD:【CSCD2017_2018】；

基金：国家自然科学基金(U1504613);河南省高校科技创新团队计划(17IRTSTHN009)

语种：中文

中文关键词：无线网络室内定位;压缩感知算法;训练样本;聚类分析;和声搜索算法;神经网络

外文关键词：wireless indoor positioning; compressed sensing algorithm; training samples; mean clustering analysis; harmony search algorithm ; neural network

摘要：室内环境复杂多变,无线信号具有强烈的时变性,支持向量机存在定位效率低,神经网络参数难以确定等难题。为了改善无线网络室内的定位效果,提出了和声搜索算法优化神经网络的无线网络室内定位模型。首先收集无线网络定位的训练样本,采用压缩感知算法减少训练样本的规模,然后采用聚类算法对样本进行聚类分析,选择最有效的训练样本,最后采用和声搜索算法优化神经网络实现无线网络定位,并通过具体仿真对比实验测试了该算法的可行性。测试结果表明,该算法的定位效果可以满足无线网络的定位实际要求。
Indoor environment is complex and changeable,and wireless signal has strong time- varying. Support vector machine has low positioning efficiency while neural network is difficult to determine the parameter. In order to improve the positioning performance in wireless network, a novel wireless positioning algorithm based on harmony search algorithm optimizing neural network is proposed. Firstly, training samples of wireless network are collected and the size of training samples is reduced by a compressed sensing algorithm ; secondly, clustering algorithm is used to cluster the samples; finally, harmony search algorithm is used to optimize neural network and feasibility is tested by simulation experiments. Test results show that the positioning results of the proposed algorithm can meet the actual requirements of wireless network positioning.