噪声干扰下的传感网络敏感信号挖掘模型

Sensitive Signal Mining Model for Sensor Network Under Noise Interference

传统的传感网络信号挖掘方法在噪声干扰下,以降低波动信号参与通信为代价调控网络平稳度,极大降低了网络信号传递效率,存在较大的弊端。提出一种基于改进中值滤波的神经网络敏感信号挖掘方法,分析噪声干扰下无线传感网络信号特征,采用改进中值滤波的神经网络对原始含噪敏感信号进行中值滤波,利用中值滤波抑制噪声干扰产生的敏感信号值,通过BP神经网络去除敏感信号中的噪声,采用梯度下降方法在信号权矢量空间中求取误差函数的极小值,获取使误差函数极小化的权值组合,也就是待挖掘的传感网络敏感信号最佳解,实现传感网络敏感信号的准确挖掘。实验结果表明,所提方法能更好地过滤噪声,有效挖掘出敏感信号,具有较高的鲁棒性和自适应特性。

Traditional sensor network signal mining methods under the noise interference, in order toreduce the fluctuation signal in smooth communication expense control network, greatly reduce thenetwork signal transmission efficiency, there is a big drawbacks. In this paper, a neural network based onimproved median filtering sensitive signal mining method, to analyze the causes of noise and feature,with the improved median filtering of the neural network to the original sensitive signal with noise,median filtering by using median filter to suppress noise sensitive signal value, eliminate the noise in thesensitive signal by BP neural network and gradient descent method is used in signal power vector spaceto calculate the minimum error function, obtain the error function minimization weight combination,namely＇s sensor is sensitive to network signal optimal solution mining, sensor network sensitive signalaccurate mining. The experimental results show that the proposed method can filter noise better, dig upsensitive signals effectively, has high robustness and adaptive features.