This technology identifies potential indoor leaks by detecting abnormal patterns in water usage data periodically collected from remote water meters. It converts daily and weekly usage into one-dimensional data using principal component analysis and automatically determines the presence of a leak by comparing statistical values against Grubbs' critical values.
Users often struggle to detect indoor leaks caused by pipe bursts, open faucets, or toilet malfunctions, even when water consumption increases significantly. Even when meter readers flag a spike in usage, it is difficult to distinguish between a genuine leak and a simple increase in consumption, creating a need for a method that can accurately identify leaks using data alone without requiring on-site visits.
This technology flags days where the minimum water usage is non-zero as abnormal, integrates daily and weekly usage into a single metric by projecting them onto the maximum variance axis, and identifies a leak if the value exceeds the 5% significance level threshold. Based on statistical outlier testing, it offers high accuracy in distinguishing between temporary usage spikes and persistent leaks. It can be integrated into municipal AMI platforms and apartment management systems, enabling early detection of leaks without site visits, thereby reducing water billing disputes and resource waste.
This invention was developed with support from the Small and Medium Business Administration for the development of an IoT-based integrated management program for ultrasonic water meters, as well as long-life piezoelectric ceramic materials and flow sensor modules.
N/A
