物联网感知层传感节点故障诊断研究∗

摘要

In view of Internet of Things networking perception layer node fault diagnosis problem, this arti-cle put forward to WSNs node fault diagnosis method of optimize the RBF neural network based on Ant Col-ony ClusteringAlgorithm, from the node hardware module failure and node failure rate influence on the accu-racy of fault diagnosis of the two aspects in the WSNs node fault diagnosis. The improved ant colony optimi-zation RBF neural network of initial weights is applied to the fault diagnosis of the WSNs node. Parallel op-timization characteristics of ant colony algorithm and adaptive adjustment coefficient of volatile characteris-tics as the clustering algorithm to determine the initial weights of RBF neural network, at the same time a-dopt crop reduction RBF neural network hidden layer, optimize network structure. Through the experimental results show that the optimization of RBF neural network based on Ant Colony ClusteringAlgorithm of WSNs node fault diagnosis method can accurately achieve perception of fault diagnosis, compared with other meth-od has higher diagnosis accuracy.%针对物联网感知层节点故障诊断问题，提出基于蚁群聚类优化RBF神经网络的WSNs节点故障诊断算法。将从节点硬件模块故障和节点故障率对故障诊断精度的影响两个方面研究WSNs节点故障诊断。将改进的蚁群聚类优化RBF神经网络的初始权值应用到WSNs节点故障诊断研究中。利用蚁群算法并行寻优特性和自适应调整挥发系数特征作为聚类算法来确定RBF神经网络初始权值，同时采用裁剪约简RBF神经网络隐含层、优化网络结构。通过实验结果表明，基于蚁群聚类优化RBF神经网络的WSNs节点故障诊断方法能准确实现感知节点的故障诊断，与其它方法相比具有更高的诊断精度。