基于时间序列和神经网络的温室传感器节点故障诊断

温室无线传感器网络中故障节点会产生并传输错误数据,不仅消耗节点的能量和带宽,而且导致错误决策。针对此问题研究一种准确判断节点故障状态的方法。采用时序分析和遗传BP神经网络,建立基于时间序列和神经网络的传感器节点故障诊断系统,通过对传感器样本数据进行时序分析,提取模型参数作为特征向量,并以此对遗传BP神经网络进行网络训练,实现传感器节点故障的诊断。试验结果表明：该方法能够有效地识别传感器节点故障类型,15组测试样本的输出矢量与同类故障基准矢量的欧式距离和为0.007,识别正确率为100%。

Fault diagnosis of greenhouse sensors nodes based on timeseries and neural network

The existence of faulty sensor nodes in wireless sensor networks (WSN) causes not only degradation of the network quality of service,but also a huge burden of the limited energy.Therefore,a method of accurate assessment about node failure status needs to be discussed.This paper adopts time series analysis and GA-BP neural network to establish a sensor nodes fault diagnosis system based on the combination of time series analysis and GA-BP neural network.With the aid of analyzing the sensor sample data and extracting model parameters as characteristic vectors,a net training process based on characteristic vector was used to build a GA-BP neural network to detect the sensor nodes faults and improve the diagnosis accuracy.The research result shows that the fault diagnosis method can effectively detect the sensor nodes failure type,the Euclidian Distance Sum of 15 test samples output vector and similar failure vector is 0.007 and the average recognition success rate is 100%.