Online leakage detection system based on EWMA-enhanced Tukey method for water distribution systems. Issue 1 (30th December 2022)

• Record Type: Journal Article
• Authors: Wan, Xi; Farmani, Raziyeh; Keedwell, Edward
• Is Part Of: Journal of hydroinformatics. Volume 25:Issue 1(2023)
• Page Start: 51
• Page End: 69
• Publication Date: 2022-12-30
• Subjects: data-driven methods -- leakage detection -- online detection -- statistical methods -- water distribution systems; Hydrology -- Data processing -- Periodicals; Geographic information systems -- Periodicals; Geographic information systems; Hydrology -- Data processing; Electronic journals; Periodicals; 551.480285
• DOI: 10.2166/hydro.2022.079
• Languages: English
• ISSNs: 1464-7141

Abstract

Abstract: Real-time leakage detection based on pressure and flow data has become increasingly essential for water distribution systems (WDSs). Recent data-driven leakage detection approaches have largely focused on burst detection characterised as sudden outflow or sudden pressure drops but did not mention the ability to detect gradual leakage events that do not have sudden change and could cause more water loss. This study proposes an online leakage detection system based on the exponential weighted moving average (EWMA)-enhanced Tukey method to help monitor gradual leakage events of WDSs. The proposed online system comprises three main parts: data pre-processing, the online detection sub-system, and the parameter updating sub-system. The proposed online system is based on lightweight and powerful statistical tools without complex model construction. The effectiveness of the proposed system is demonstrated on leakage datasets under various real-world scenarios, including gradual leakages and bursts. The results showed that the proposed EWMA-enhanced Tukey method could detect gradual leakage events quickly while generating low false alarms. The proposed method is computationally effective and able to deal with non-stationary behaviours automatically. HIGHLIGHTS: This paper proposed a leakage detection method that focused on gradual leakage events. The proposed leakage detection method could adapt to the time-varying characteristics of flow monitoring data, such as the demand variation caused by weather-related issues. This paper proposed a method that is robust to the data noises and could successfully detect leakage events without generating many false alarms. Graphical Abstract