基于深度学习算法的无线传感器异常数据检测

提出了基于深度学习的异常数据检测的方法，精准检测到无线传感器异常数据并直观展现检测结果。基于无线传感器网络模型分簇原理，通过异常数据驱动的簇内数据融合机制，去除无线传感器网络中的无效数据，获取无线传感器网络有效数据融合结果。构建了具有4层隐含层的深度卷积神经网络，将预处理后的无线传感器网络数据作为模型输入，通过隐含层完成数据特征提取和映射后，由输出层输出异常数据检测结果。实验证明：该方法可有效融合不同类型数据，且网络节点平均能耗较低；包含4层隐含层的深度卷积神经网络平均分类精度高达98.44%,1000次迭代后隐含层的训练损失均趋于0,可实现无线传感器异常数据实时、直观、准确检测。

Abnormal Data Detection of Wireless Sensor Based on Deep Learning Algorithm

The abnormal data detection method based on deep learning is proposed to accurately detect the abnormal data of wireless sensors and visually display the detection results. Based on the clustering principle of wireless sensor network (WSN) model, the invalid data in WSN is removed through the data fusion mechanism driven by abnormal data, and the effective data fusion results of WSN are obtained. A deep convolutional neural network with four hidden layers is constructed, and the preprocessed wireless sensor network data is used as the model input. After data feature extraction and mapping are completed through the hidden layer, abnormal data detection results are output by the output layer. Experimental results show that this method can effectively fuse different types of data, and the average energy consumption of network nodes is low. The deep convolutional neural network with four hidden layers has an average classification accuracy of 98.44%, and the training loss of hidden layers tends to 0 after 1000 iterations, which can realize real-time, intuitive and accurate detection of wireless sensor abnormal data.