Tracking Crack Development in Smart Water Networks Using IoT Acoustic Sensors

Abstract

Internet of Things (IoT) technologies with distributed wireless sensors have been increasingly adopted in water utilities to build smart water networks (SWNs) for monitoring purposes. Based on the daily data collected from an accelerometer-based SWN, this paper proposes a new data analytic approach to detect developing cracks in water networks. The daily signals over a continuous period (e.g., 100 days) have been converted to a time-frequency power spectral density (PSD) heatmap. An analytic approach to detect developing cracks on the PSD heatmap has been formulated using a Spearman’s rank correlation coefficient. With flexible temporal window lengths and frequency-associated weights adopted, the method involves an optimal search concept in the time-frequency domain for evidence of developing cracks. The implementation of the new method to field data collected from an SWN illustrates that the method can robustly detect developing cracks at their incipient stage, and thus allow adequate time for proactive repair before evolving into pipe breaks. The method is tolerant of noise, which is commonly present in the data collected by the sensors deployed in city areas.