基于神经网络的无线网络故障节点定位技术

Wireless network fault node location technology based on neural network

传统无线网络故障节点定位方法无法有效处理节点功率波动以及模糊环境对故障节点定位精度的干扰。提出基于小波神经网络的无线网络故障节点定位方法,分析了小波神经网络在节点故障定位的三种作用形式,融合形式1和3对冗余节点故障进行定位,将小波神经网络当成预测器,将前一采样时刻的正常输出交叉输入n个小波神经网络,获取节点当前时刻的预测输出值,取节点预测输出值和真实输出值的残差,若该残差值高于阈值,则说明该节点是故障节点。实验结果表明,所提故障节点定位方法能够对节点的附加、倍数以及短路故障进行准确定位。

Since the traditional wireless network fault node positioning method is unable to effectively handle node power fluctuation and the fuzzy environment disturbance to the fault node location accuracy, a wireless network fault node localization method based on wavelet neural network is proposed. Three kinds of effect forms of wavelet neural network in the node fault location is analyzed. The fusion forms 1 and 3 are used to locate the redundant node failure. The wavelet neural network is taken as a predictor. The normal output of the previous sampling time is input into n wavelet neural networks in cross form to get node prediction output value of the current time. The residual error between the node prediction output value and actual output value is deduced. If the residual error value is higher than the threshold, it means that the node is fault node. The experimental results indicate that the proposed fault node positioning method can accurately position the node with additional, multiple and short trouble faults.