基于优化神经网络与聚类的加权定位算法  

A weighted localization algorithm based on optimized neural network and clustering

作　　者：李轶 LI Yi(The 716 Research Institute of CSSC,Lianyungang 222000,Jiangsu Province,China)

机构地区：[1]中国船舶集团公司第七一六研究所,江苏连云港222000

出　　处：《信息技术》2023年第5期13-18,24,共7页Information Technology

基　　金：工业和信息化部2019年工业互联网创新发展工程——工业企业网络安全综合防护平台项目(TC190H3WQ)。

摘　　要：针对RSSI测距误差较大的现象,为进一步提高加权质心定位算法精度,提出了一种基于MEA-BP神经网络与模糊C均值聚类的加权定位算法。该算法首先引入思维进化算法与BP神经网络以优化测距信息;采用三边测量法得到多个初始定位结果并组成集合,经模糊C均值聚类处理后,使用加权质心定位算法得到每个聚类簇针对于未知节点的估计位置;将每个聚类簇中数据点的数量视为权值,再次使用加权质心定位算法,得到未知节点的最终定位坐标。仿真结果表明,该算法显著降低了二维场景中无线传感器网络定位误差。In order to overcome large error of RSSI,improve the accuracy of weighted centroid localization algorithm,a weighted localization algorithm based on MEA-BP neural network and fuzzy C-means clustering is proposed.This algorithm uses MEA-BP neural network model to optimize ranging information firstly,then uses trilateral measurement method to get multiple initial positioning results about unknown node and form a set.After clustering the results by fuzzy C-means,the estimated position of unknown node in each cluster is obtained by using the weighted centroid localization algorithm.Nextly,the number of data points in each cluster is regarded as weight,the weighted centroid positioning algorithm is used to get the final positioning coordinates of unknown node.Simulation results show that the algorithm improves the positioning accuracy of wireless sensor network in two-dimensional scene significantly.

关 键 词：无线传感器网络 BP神经网络 思维进化算法 模糊C均值聚类 加权质心定位算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]