基于贝叶斯和层次模型的传感器网络节点故障预测研究  

Study on fault prognosis of sensor network nodes using Bayesian and hierarchical model

作　　者：何永强[1] 宫玉荣[2] 朱予聪[1] 

机构地区：[1]河南工程学院计算机学院,河南郑州451191 [2]郑州成功财经学院共同学科部,河南郑州451200

出　　处：《河南工程学院学报（自然科学版）》2015年第4期63-68,共6页Journal of Henan University of Engineering：Natural Science Edition

基　　金：河南省科技厅科技计划课题(152102210027);河南省高等学校重点科研项目(15A520054)

摘　　要：提出了运用贝叶斯理论和层次模型对传感器网络节点故障进行预测的方法,结合传感器网络各节点处理信息的时序关系,用定性分析将时间信息融合到节点中分析故障传播机制和故障预测.根据节点的历史信息和当前运行情况,预测和确定节点的故障概率;利用参数学习和概率推理,预测上层节点的故障概率.通过仿真实验并与其他预测方法进行比较分析,验证了该方法的可靠性和精确性,为传感器网络节点的故障预测提供了新的思路和方法.This paper presents the method for sensor network node failure prediction with the relationship between timing and level using Bayesian model methods. The method combines processing information in each node sensor network and uses qualitative analysis to integrate time information into the node to analyze the fault propagation mechanism and failure prediction. According to the probabili- ty of failure history information of node and the current operating conditions, nodes failure probabilities are forecasted and determined. Using the parameters of learning and probabilistic reasoning, the probability of failure of the upper node is predicted. Through simula- tion analysis and comparative analysis with other forecasting methods, the accuracy and precision of the method for sensor network node failure prediction is verified, providing new ideas and methods for nodes failure detection.

关 键 词：贝叶斯 层次模型 传感器网络 故障预测 模糊隶属函数 定性趋势分析 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]