Noise Removal for the Steady-State Pressure Measurements Based on Domain Knowledge of Water Distribution Systems

Abstract

The pressure sensors have been widely used to capture the steady-state pressure data for the management of water distribution systems (WDSs). The background noise pollution in WDSs can lead to signal degradation, reducing the reliability of sensor data. Although numerous algorithms have been developed for noise removal, their optimal parameter selection is problem-dependent. The steady-state pressure sensor data are often linearly correlated, and this correlation is related to hydraulic distance. This is a key feature of WDS pressure sensor data that should be considered in noise-removal algorithms. Based on this domain knowledge, two noise-removal metrics are proposed to evaluate the performance of the noise-removal algorithm and aid in selecting the optimal parameters. In addition, four widely used noise-removal algorithms have been reviewed and used for removing noise from the pressure sensor data. The results demonstrate that the noise-removal algorithm can efficiently remove potential random noise and enhance the reliability of the sensor data.