欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2022, Vol. 39 ›› Issue (2): 232-239.DOI: 10.7523/j.ucas.2020.0004

• 电子信息与计算机科学 • 上一篇    

基于改进灰狼算法优化BP神经网络的无线传感器网络数据融合算法

曹轲1,2, 谭冲1, 刘洪1, 郑敏1   

  1. 1 上海微系统与信息技术研究所 中国科学院无线传感网与通信重点实验室, 上海 200050;
    2 中国科学院大学, 北京 100049
  • 收稿日期:2020-01-06 修回日期:2020-04-08 发布日期:2021-05-31
  • 通讯作者: 曹轲
  • 基金资助:
    中国科学院青年创新促进会(2018269)和上海市科委科研计划项目(18511106400)资助

Data fusion algorithm of wireless sensor network based on BP neural network optimized by improved grey wolf optimizer

CAO Ke1,2, TAN Chong1, LIU Hong1, ZHENG Min1   

  1. 1 Key Laboratory of Wireless Sensor Networks and Communications of CAS, Shanghai Institute of Microsystem and Information Technology, Shanghai 200050, China;
    2 University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2020-01-06 Revised:2020-04-08 Published:2021-05-31

摘要: 为提高无线传感器网络数据融合精度,降低网络能耗,延长网络生存时间,提出基于改进灰狼算法优化BP神经网络的无线传感器网络数据融合算法(IGWOBPDA)。首先为平衡灰狼算法全局与局部搜索能力提出改进控制参数和动态权重更新位置的改进灰狼方案,利用改进灰狼算法对BP神经网络初始阈值和初始权值进行优化以解决数据融合中BP神经网络对初值敏感、易陷入局部最优的问题;其次考虑到无线传感网实际传输节点能耗和分簇情况,提出基于节点剩余能量参数和节点密度参数的分簇方案,通过调整参数的权重因子来适应网络数据融合传输过程中的实际情况。仿真实验结果表明,对比BPNDA算法和GAPSOBP算法,IGWOBPDA算法在不同数据集下有更好的数据融合精度和更快的收敛速度,并且能有效地减少数据传输量,降低节点能耗,延长网络生存时间,在不同网络规模下保持稳定性。

关键词: 无线传感器网络, 数据融合, BP神经网络, 灰狼算法, 控制参数, 权重因子

Abstract: In order to improve the accuracy of the fusion data in wireless sensor network, reduce the energy consumption, and extend the network lifetime, data fusion algorithm of wireless sensor network based on improved grey wolf optimizer to optimize BP neural network (IGWOBPDA) is proposed in this paper. Firstly to balance the global and local search capabilties, the improved considering the actual energy consumption and clustering of the wireless sensor network's transmitting nodes, a clustering scheme based on node residual energy parameters and node density parameters is proposed,which adjusts the weighting factors to adapt to the actual situation of the network data fusion transmission process. Compared with BPNDA algorithm and GAPSOBP algorithm, simulation results show that IGWOBPDA algorithm has better data fusion accuracy and convergence in different data sets, which can effectively reduce the amount of data transmission and node energy consumption, extend network survival time, and maintain stability under different network scales. control parameter and the method of dynamic weight update position is proposed in the paper, which aims at the problems that the initial value of BP neural network in wireless sensor network data fusion algorithm is sensitive and the result can easily be the local optimal solution. Secondly,

Key words: wireless sensor network, data fusion algorithm, BP neural network, grey wolf optimizer, control parameter, weighting factor

中图分类号: