摘要
提出了一种将粗糙集与RBF神经网络相结合的故障诊断技术应用于无线手持器,对无线传感器网络节点进行故障诊断,并将RSRBF与常规RBF、RSBP进行比较。首先用粗糙集中的约简算法对WSN中节点故障诊断信息进行约简,然后使用训练样本对神经网络进行训练,最后,使用训练后的神经网络对测试样本进行诊断。在仿真试验中,通过与常规RBF、RSBP比较,结果表明RSRBF网络的训练速度远高于RSBP网络,且比常规RBF具有更高的故障诊断准确率。
This paper proposes a fault diagnosis which combines rough sets with RBF neural network to diagnose nodes in WSNS. It compares RSRBF with conventional RBF and RSBP. Firstly,we use the rough set reduction algorithm to reduce the node fault diagnosis information in WSN. Then,training samples are used to train the neural networks. Finally,the neural network is used to diagnose the test samples. In the simulation,through the comparison with the conventional RBF and RSBP networks,the RSRBF shows higher training speed than RSBP and higher accuracy of fault diagnosis than conventional RBF.
出处
《仪表技术》
2015年第11期5-9 46,46,共6页
Instrumentation Technology
