Water leak detection using FIDO AI

Adasa has carried out a pilot project in the town of Anna that has analysed a total length of 16.5 km of water distribution network by implementing disruptive digital solutions such as artificial intelligence (AI) based on FIDO Tech technology.

The project has deployed a total of 210 acoustic sensors (bugs) in one week, whose information has enabled a global mapping of valuable information that locates areas free of leaks and critical areas.

FIDO Tech applies the machine learning algorithm in several phases:

1. High sensitivity and accuracy in detection, even at very low vibration levels that produce many large volume leaks.

2. Acoustic data collection, data transmission (LwM2M, https and COAPs) to the Microsoft Azure cloud server and automatic analysis. Apply proprietary data encryption and compression and cybersecurity mechanisms between bug and app and between bug and server (OAuth2 and MFA).

3.  Filters out noises caused by other urban sources such as traffic, valves, pumps, etc., which can generate false alarms.

In the localisation phase, we have precisely delimited the leaks and established evidence points for excavation.

The project developed by Adasa and Aqlara Ciclo Integral del Agua has helped Anna Town Council to:

1. Detect a significant number of water leaks following the identification of possible locations using correlation and field marking techniques.

2.   Precisely locate the point of water leakage.

3.   Optimise the analysis of areas likely to contain potential water leaks by clustering the detected points.

4.   Classify the size of water leaks as small, medium or large.

5.   To rule out possible problem areas in the distribution network, after having obtained the result of a leak-free zone.

6.   Discover existing problems in different network fittings, e.g. valves and fire hydrants that were not in good working order.