无线传感器网络自适应预测加权数据融合算法  

Adaptive Forecast Weighting Data Fusion Algorithm for Wireless Sensor Network

作　　者：余修武[1,2] 范飞生 周利兴 张枫[1,2] 

机构地区：[1]南华大学环境保护与安全工程学院,湖南衡阳421001 [2]金属矿山安全与健康国家重点实验室,安徽马鞍山243000

出　　处：《传感技术学报》2017年第5期772-776,共5页Chinese Journal of Sensors and Actuators

基　　金：金属矿山安全与健康国家重点实验室开放基金项目(2016JSKSSYS04);湖南省重点研发项目(2015SK2005);湖南省教育厅科研重点项目(15A161);江西省自然科学基金项目(20122BAB201050)

摘　　要：为提高无线传感器网络监测系统的可靠性及寿命,提出了一种基于分簇的自适应的预测加权数据融合(AFWDF)算法。AFWDF算法依据数据在时间上的相关性,建立预测模型。源节点与簇头利用前期监测数据的变化态势自适应调整预测模型参数对后期数据进行预测,源节点通过预测值与测量值比较提取特征值和剔除异常值,簇头根据特征值和预测值还原监测值,并计算监测值可信度和权重进行加权数据融合。通过性能分析及仿真,得出AFWDF可靠性较高,且在模拟环境下网络寿命周期比SAEMDA和BPNDA算法提高了15%左右。In order to improve the reliability and extend lifetime of wireless sensor network （ WSN ） monitoring system ,an adaptive forecast weighting data fusion （AFWDF）algorithm based on clustering is proposed. AFWDF builds a forecast model based on the temporal correlation of data.The source node and the cluster head adaptively adjust the prediction model parameters by using the change trend of the early monitoring data to predict the later da- ta.The source node extracts the eigenvalues and eliminates the abnormal values by comparing the predicted values with the measured values. The cluster head restores the monitored values and calculates the reliability and weight of the monitored values to fuse the them. Through the performance analysis and simulation, it is concluded that the AFWDF algorithm has high reliability,and the network life cycle is about 15% higher than SAEMDA and BPNDA in the simulation environment.

关 键 词：无线传感器网络 数据融合 自适应预测模型 环境监测 可信度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]