摘要
节点故障诊断是无线传感器网络持续性监测服务的关键步骤。为准确高效地得到诊断结果,提出一种基于数据驱动的故障诊断算法。对每个节点获取的信息建立空间高维向量,引入自身历史数据和邻居节点数据构造十字滑动窗口,并赋予十字方向上自定义的故障权重,通过检测异常向量达到故障诊断的目的。实验结果表明,该算法能够克服计算量大、故障判定条件苛刻等不足,与分布式故障诊断算法相比,可使故障诊断正确率提高15.5%,故障误警率下降4.89%。
Node fault diagnosis is the key step in the continuous monitoring service of Wireless Sensor Network( WSN).In order to implement accurate and efficient node fault diagnosis,this paper proposes a fault diagnosis algorithm based on data driven. It builds high-dimensional vector in space through the information from each node, establishes cross-sliding window by its historical data and neighbor node data,gives customized weights of fault in the cross direction,and ultimately achieves the goal of fault diagnosis by detecting abnormal vector. Experimental results show that this algorithm can overcome defects such as large amount of calculation,harsh conditions of fault judgement,etc. Compared with Distributed Fault Diagnosis( DFD) algorithm,it improves Fault Diagnosis Accuracy( FDA) by 15.5% and reduces False Alarm Rate( FAR) by 4.89%.
出处
《计算机工程》
CAS
CSCD
北大核心
2017年第9期105-109,共5页
Computer Engineering
基金
国家发改委项目"森林监测无线传感器网络管理与事件检测技术研究"(Q5025001201502)
关键词
无线传感器网络
节点故障诊断
数据驱动
向量空间模型
十字滑动窗口
故障权重
Wireless Sensor Network(WSN)
node fault diagnosis
data driven
Vector Space Model(VSM)
cross-sliding window
weight of fault