一种基于相关分析的多元回归数据估计方法  

Data evaluation method using multiple regression based on correlation analysis

作　　者：焉晓贞 谢红[1] 王桐[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,哈尔滨150001

出　　处：《沈阳工业大学学报》2013年第2期212-217,共6页Journal of Shenyang University of Technology

基　　金：中国博士后科学基金资助项目(20080440840);中央高校基本科研业务费专项基金资助项目(HEUCF100826)

摘　　要：针对无线传感器网络中数据估计方法存在输入变量较多、估计计算复杂度较高和估计效率低等缺点,提出了基于相关分析的多元回归估计方法.对无线传感器网络监测的数据序列进行相关分析,找出与当前数据相关性较强的其他历史监测数据,采用这些历史监测数据进行多元回归建模和估计.在保证估计精度的前提下,降低估计的计算复杂度,提高无线传感器网络中缺失监测数据估计的效率.实际无线传感器网络采集数据的实验分析结果表明,该方法具有较高的估计效率和较小的估计误差,能够有效地估计无线传感器网络中缺失的传感数据,具有一定的应用价值.Aiming at such disadvantages as many input variables,high computational complexity and low evaluation efficiency existing in data evaluation methods for wireless sensor networks,a multiple regression evaluation method based on correlation analysis was proposed.The correlation analysis on the data sequence monitored by wireless sensor networks was carried out,and other historical monitoring data with higher correlation to current data were found out.The obtained historical monitoring data were used to perform the multiple regression modeling and evaluation.On the precondition of ensuring the evaluation accuracy,the computational complexity of evaluation was reduced,and the evaluation efficiency of missing monitoring data in wireless sensor networks was enhanced.The analysis results on the data collected by wireless sensor networks show that the proposed method has higher evaluation efficiency and lower evaluation error,can effectively evaluate the missing sensor data in wireless sensor networks,and has certain application value.

关 键 词：无线传感器网络 数据缺失 数据估计 多元回归 相关分析 最小二乘 支持向量机 支持向量回归 实际传感数据 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]