Adriático: a new system to combat water leakage

A team, including researchers from the CVBLab-HumanTech group at the Universitat Politècnica de València (UPV), has developed a new intelligent system for monitoring water networks. The system enables real-time detection and location of leaks, infiltrations, intrusions, or structural damage in pipelines.

Developed within the framework of the ADRIATICO project, the system combines distributed fibre-optic sensing with artificial intelligence models and edge computing techniques. The integration of these technologies enables continuous analysis of large volumes of data directly within the infrastructure, without the need to send data to central servers.

“The system uses signals generated by fibre?optic sensors installed along the water network. Based on these data, deep?learning models identify patterns associated with different types of events, enabling accurate and automated monitoring directly on the devices deployed in the network,” explains Adrián Colomer, researcher at UPV’s CVBLab?HumanTech group.

The system can be applied to leak detection in water supply networks, monitoring of sanitation systems, intrusion control in critical infrastructures and predictive maintenance of hydraulic installations. All of this contributes to optimising the integrated management of the water cycle in urban and agricultural environments. Key advantages include continuous monitoring of tens of kilometres of pipelines using a single distributed sensing system, reduced maintenance costs, high-accuracy anomaly detection, and improved sustainability through reduced water losses.

This development has been made possible by a multidisciplinary technical team with extensive expertise in artificial intelligence, data analysis, and high-precision fibre-optic sensing, bringing together the CVBLab-HumanTech group and the companies Fibsen and Artikode Intelligence.

Precise localisation

One of the project’s main outcomes is the ability of the developed models to detect and accurately locate anomalies, leveraging the high sensitivity and precision of fibre-optic sensors to pinpoint the exact point in the pipeline where the problem occurs. The models are also able to distinguish between different types of events — such as leaks, external vibrations or tampering — and to reduce the number of false alarms.

“This is essential for effective management of incidents and makes it possible to prioritise intervention in the most critical cases, such as water leaks. In addition, real?time analysis can be carried out directly on site, reducing the volume of transmitted data and improving the efficiency and scalability of the system in large infrastructures,” highlights Luis Romero, CTO of Fibsen.

The system has been validated in real environments and can be adapted to different conditions, such as variations in terrain, pipe diameter or network configuration. “This flexibility facilitates integration into existing infrastructures. Our next objective is validation under real operational conditions and the integration of the technology into commercial solutions,” concludes Ana Martín, CEO of Artikode Intelligence.