基于
改进
BP
神经网络
无线
传感
网络
数据
融合
方法
信息通信基于改进BP神经网络的无线传感网络数据融合方法(兰州城市学院实验室与设备管理中心,甘肃兰州7 30 0 7 0)摘要:为提高融合数据与网络节点的适配度,引进改进BP神经网络,以无线传感网络为例,设计一种全新的数据融合方法。采集无线传感网络数据,引进Kalman滤波器,根据滤波器在空间中的线性表达区域,给定一个无线传感网络数据的线性归回范围,设计无线传感网络数据的预处理与特征提取;利用改进BP神经网络的数据反向传播特点,参照网络在学习中的映射规则,调节网络传播节点的阈值与权值训练数据;设定无线传感网络数据集合的参考视场和步长,计算数据与节点适应值;选用Max-min标准化处理法对数据归一化处理,采用LEACH算法深度融合网络数据。设计对比实验,实验结果证明:该方法可在提高融合数据与节点适应值的基础上,提高节点对数据的收敛速度,以此达到最优的融合效果。关键词:改进BP神经网络;归一化处理;适应值;融合方法;数据;无线传感网络中图分类号:TP3930引言为提高网络数据的生存时间与生存能力,有关单位提出收稿日期:2 0 2 3-0 1-0 5作者简介:张红蕾(198 9-),女,甘肃兰州人,硕士,讲师,研究方向:数据挖掘。4 Chen,Liang-Chieh,et al.Encoder-decoder with atrous separ-able convolution for semantic image segmentation.Proceedingsof the European conference on computer vision(ECCV).2018.5 Barlow,J.,S.Franklin,and Y.Martin,High spatial resolutionsatellite imagery,DEM derivatives,and image segmentation forthe detection of mass wasting processes.Photogrammetric En-gineering&Remote Sensing,2006.72(6):p.687-692-687-692.6杜建军,李大壮,廖生进,等.基于 CT图像和 RAUNet-3D 的玉米籽粒三维结构测量.农业机械学报,2 0 2 2,53(12):2 44-2 53+2 8 9.7王大方,刘磊,曹江,等.基于空洞空间池化金字塔的自动驾驶图像语义分割方法.汽车工程,2 0 2 2,44(12):18 18-18 2 4.DOI:10.19562/j.chinasae.qcgc.2022.12.003.8 Chen,X.,et al.Learning active contour models for medical im-age segmentation,in Proceedings of the IEEE/CVF conferenceon computer vision and pattern recognition.9 Dai,W,et al.,Structure correcting adversarial network for chest X-rays organ segmentation.2020,Google Patents.10 Fu J,Liu J,Tian H,et al.Dual ttention network for scene seg-mentation C/Proceedings of the IEEE/CVF Conference onComputer Vision and Pattern Recognition.2019:3146-3154.11 Multi-Scale Context Aggregation by Dilated Convolutions.J.Fisher Yu;Vladlen Koltun.CoRR.201512郑帅.语义分割技术在船舶卫星图像识别中的应用J.舰船科学技术,2 0 2 2,44(14):155-158.13 Jegou,S.,et al.The one hundred layers tiramisu:Fully con-volutional densenets for semantic segmentation.in Proceed-ings of the IEEE conference on computer vision and patternrecognition workshops.14 Long,J.,E.Shelhamer,and T.Darrell.Fully convolutionalnetworks for semantic segmentation.in Proceedings of theIEEE conference on computer vision and pattern recognition.15 Hung,W.-C.,t al.,Adversarial learning for semi-supervised semantic segmentation.arXiv preprint arXiv:1802.07934,2018.16 Wu,Q.,et al.,Fully convolutional networks semantic seg-mentation based on conditional random field optimization.Journal of Computational Methods in Sciences and Engin-eering,2021.21(5):p.1405-1415-1405-1415.772023年第0 5期(总第 2 45 期)张红蕾文献标识码:A文章编号:2 0 96-97 59(2 0 2 3)0 5-0 0 7 7-0 3了针对无线传感网络数据的融合处理方案,通过对多节点数据的融合处理,掌握不同类别数据的特征,降低单链路数据传一一一+一+一+17 Ronneberger,O.,P.Fischer,and T.Brox.U-net:Convolu-tional networks for biomedical image segmentation.in Inter-national Conference on Medical image computing and com-puter-assisted intervention.18 Hong,S.,H.Noh,and B.Han,Decoupled deep neural net-work for semi-supervised semantic segmentation.Advancesin neural information processing systems,2015.28.19 Scudder,H.,Probability of error of some adaptive pattern-recognition machines.IEEE Transactions on InformationTheory,1965.11(3):p.363-371-363-371.20 Zhan,X.,et al.Mix-and-match tuning for self-supervised sem-antic segmentation.in Proceedings of the AAAI Conferenceon Artificial Intelligence.2 Girshick,R.,et al.Rich feature hierarchies for accurate objectdetection and semantic segmentation.in Proceedings of theIEEE conference on computer vision and pattern recognition.22 Chen,X.and A.Gupta,An implementation of faster rcnn with study for region sampling.arXiv preprint arXiv:1702.02138,2017.23 He,K.,et al.Mask r-cnn.in Proceedings of the IEEE inter-national conference on computer vision.24 Wei,Y.,et al.,Stc:A simple to complex framework for wea-kly-supervised semantic segmentation.IEEE transactions onpattern analysis and machine intelligence,2016.39(11):p.2314-2320-2314-2320.25 Wei,Y.,et al.,CNN:Single-label to multi-label.arXiv pre-print arXiv:1406.5726,2014.26 Goodfellow,I.,et al.,Generative adversarial networks.Com-munications of the ACM,2020.63(11):p.139-144-139-144.27 Radford,A.,L.Metz,and S.Chintala,Unsupervised repres-entation learning with deep convolutional generative adver-sarial networks.arXiv preprint arXiv:1511.06434,2015.28 Denton,E.L.,et al.,Deep generative image models using alaplacian pyramid of adversarial networks.Advances in neu-ral information processing systems,2015.28.29 Pathak,D.,et al.Context encoders:Feature learning by inpa-inting.in Proceedings of the IEEE conference on computervision and pattern recognition.Changjiang Information&Communications输量,减少无线传感器网络能耗,提高节点数据的生存时间。目前,现有的研究成果大多集中在以下两个方面,其一为通过改善无线传感器网络负载,提升数据的传输能力;其二为通过提高无线传感器网络数据处理精度,完善节点运行环境。尽管目前部分学者已经取得了阶段性的研究成果,但要发挥网络节点数据传输的价值与效能,应在现有工作的基础上,结合相关工作的具体需求,对现有方法进行优化,通过此种方式避免数据在传输中出现丢失、异常等现象2。为此,本文将在此次研究中,引进改进BP神经网络,以无线传感网络为例,设计一种全新的数据融合方法。1无线传感网络数据预处理与特征提取为实现对无线传感网络数据的融合,在设计方法前,对无线传感网络进行预处理。在无线传感网络中,每个时间节点都会接收到大量的数据,并将数据储存在一个数据流集合执行器中,本文通过执行器中的阵列储存所收到的资料,从中提取出最有价值的一段时间,然后将其传送给路由器中的其他节点,以供完成数据预处理3-4,此过程如下图1所示。数据包数据流缓冲区数据包间隔图1无线传感网络数据预处理模块完成对数据包的预处理后,考虑到节点分布数据存在密度稀疏差异的问题,为了去除无线传