传感器网络同步态的节点故障诊断算法

首先介绍了复杂网络同步态的概念,以传感器量测数据为节点,定义了随时间动态变化的传感器网络,采用数学分析方法定量描述了传感器网络的动力学机制,给出了传感器网络同步态的数学定义、计算方法及其实际的物理含义。理论推导表明,同步态从全局角度评价传感器网络的健康程度,以量测数据距离关联性定义复杂网络的耦合矩阵A=(aij)N×N,并以该耦合矩阵零特征值对应的左特征向量(ξ1,ξ2,...,ξN)来刻画传感器网络节点的局部细节信息,进而衍生出基于传感器网络同步态的节点故障诊断算法,实现传感器网络的故障诊断。实验仿真了由100个传感器组成的复杂网络,采集了在稳定运动60s期间的的量测数据,每个量测数据长度为5 000,其中有3个传感器处于间歇增益故障状态,以此来验证基于传感器网络同步态的节点故障诊断算法的有效性。结果表明,该算法不仅可以很好地跟踪整个传感器网络的工作状态,实时监测每个传感器网络节点的故障,而且可以利用传感器网络节点故障之间的相关性有效地识别出传感器量测数据的异常是由外界量测对象的改变还是由传感器本身故障引起的。该算法为全局评估传感器网络的工作状态和监测网络节点的局部故障提供了一个新颖可行的研究思路,期望为相关领域的研究学者提供有益的参考。

Nodes fault diagnosis algorithm based on sensor network synchronous state

Absttact:First,the concept of complex network synchronous state was introduced in this paper.Then a sensor network varying with time was defined by taking the data measured by sensors as nodes,and the dynamic mechanics of the sensor network was quantitatively described with mathematical analysis method.Finally,the mathematical definition,the calculation method and the physical meaning of the sensor network synchronous state were given.The above theoretical derivation show synchronous state can globally assess the health of the sensor network.The couple matrix A= (aij )N×N of complex network was defined by the distance relevance of the measured data.And the left eigenvector (ξ1 ,ξ2 ,...,ξN ) corresponding to the zero feature of the matrix was used to character the local details of sensor network nodes.Then,a node fault diagnosis algorithm was derived based on sensor network synchronous state.A complex network which consists of 100 sensors was experimentally simulated.We collected the measured data in 60 s during the stable operation with each length of 5 000,and there were 3 sensors in intermittent gain fault state to verify the effectiveness of the proposed method.The simulation results show that the proposed method can not only track the work state of the whole sensor work and monitor the faults of each node in real-time,but also distinguish the abnormal data caused by the change of external objects or by sensor faults through combining the relevance between the node faults.The proposed algorithm can provide a feasible research idea of assessing the global state of the sensor network and monitoring the partial fault of network nodes,and it’s hoped the algorithm can provide references to researchers in related fields.