The Experiment, “Water Leakage detection”, was jointly led by University of Maribor and Mariborski Vodovod d.o.o..
Mariborski vodovod is SME located in Eastern Slovenia and is the main water supplier in that region. It has developed its own modern sensor network system for monitoring water consumption. The pipe network has been renovated for last 30 years, but due to financial limitations, some parts of pipe network are 50 years old. The monitoring consumption system provides real time data, but the data acquired by flow sensors are recorded into the database with no adequate reasoning being applied.
The goal is to monitor the consumption of water and detect possible water leakage from the pipe network, providing alarms when this happens.
The experiment will be focused on information extraction from a large scale data coming from:
- water flow meters that are already installed over the entire water pipe network to measure soil moisture;
- remote sensing data, in form of images acquired from SAR (radar satellite)
These two sources are considered as two separate experiments.
AI will make use of the large amount of data from these two sources so as to build up models allowing to efficiently estimate where water leakages are currently taking place.
AI has been used to detect and monitor water consumption. The remote sensing data can be used to detect wet zones. There exist many methods for soil moisture detection. The goal was to automatically estimate wet zones over the waterpipe network, compare the results with the water consumption of the network and build up AI based reasoning for water leakage detection using satellite remote sensing data and ground data.