基于数据压缩的WSN水质异常数据检测算法

WSN Water Quality Anomaly Detection Technology Based on Data Compression

大面积的水域通常需要部署许多传感器节点来采集数据,如果不对节点采集到的水质数据进行压缩处理,而直接进行初始数据的传输和异常检测,则会损耗传感器节点能量,降低传感器网络寿命。由于现有的压缩方法对异常数据特征的保留存在一定缺陷,因此提出一种基于小波分解的压缩感知与SVM相结合的水质异常检测模型。该模型先通过小波变换和压缩感知对初始水质数据进行二次压缩,再利用压缩处理后的水质数据训练SVM模型并进行最终的水质异常数据检测。实验结果表明,所提方法提高了大量水质数据的检测效率和水质异常数据的检测精度。

A large area of water usually needs to deploy many sensor nodes to collect data. If the water quality data collected by the nodes is not compressed, but the initial data transmission and abnormal detection are directly performed, the energy of the sensor nodes will be lost and the life of the sensor network will be reduced. Because the existing compression methods have certain defects in the retention of abnormal data features, a water quality abnormal detection model based on wavelet decomposition combined with compressed sensing and SVM is proposed. The model firstly compresses the initial water quality data through wavelet transform and compressed sensing, and then uses the compressed water quality data to train the SVM model and perform the final water quality abnormal data detection. The experimental results show that the proposed method improves the detection efficiency of a large number of water quality data and the detection accuracy of abnormal water quality data.