基于小波去噪-卡尔曼滤波的供水管网爆管检测

Pipe burst detection of water distribution system based on wavelet denoising-Kalman filter

在应用卡尔曼滤波进行爆管检测时,若受到监测误差、用水量随机波动等因素形成的背景噪音影响会导致误报率偏高,进而带来爆管检测系统鲁棒性较低的问题。鉴于此,对原始卡尔曼滤波进行了改进,首先采用小波函数对监测数据进行处理,其可以成功将周期循环信号中的有效信号及噪声信号分解至不同频域,然后使用硬阈值函数去噪后再对信号进行重构;最后利用卡尔曼滤波对重构后的信号进行爆管检测。研究结果表明:通过小波去噪,重构的信号中噪声信号占比显著降低,且信号与有效信号相似性更大;利用实际管网数据验证发现,所提算法可成功应用于实际管网爆管检测且与卡尔曼滤波相比,信号持续稳定,无剧烈波动且误报次数由6次降至0次,极大提高了爆管检测系统鲁棒性。

The original Kalman filter is improved,firstly,the wavelet function is used to process the monitoring data,which can successfully decompose the effective signal and the noise signal in the periodic cyclic signal into different frequency domains,and then the signal is reconstructed after using the hard threshold function for noise removal;finally,the reconstructed signal is used for pipe burst detection by Kalman filter.The results show that the proportion of noise signal in reconstructed signal is significantly reduced by wavelet denoising,and the signal is more similar to the effective signal.Using the actual water distribution system data verification,it is found that the algorithm proposed in this paper can be successfully applied to the actual pipe network burst detection and compared with Kalman filter ing,the signal is continuously stable,there is no violent fluctuation and the number of false alarms is reduced from 6 to 0,which greatly improves the robustness of the pipe burst detection system.