Leakmited relies on AI to protect water resources

90% yield! Leakmited’s goal is set. The Parisian start-up created in 2019 plans to exceed 90% efficiency in water distribution by reducing leaks when the current average efficiency is between 75% and 80% in France.

“This average is distorted according to the territories. Rural areas have less human and financial resources than large cities. They rather show a 65% yield”, explains Hubert Baya Toda, founder of Leakmited. To fight against the waste of water resources, communities and companies in charge of distribution can use leak detectors that they send permanently to the field to listen to the network.

“This profession shows a great shortage despite tempting offers.” The second alternative consists in equipping the network with acoustic sensors which will make it possible to have a constantly alert ear. “You need four sensors per kilometer, or around 2,000 euros per kilometer.” Not all communities have the necessary funds to invest in this direction.

AI to the rescue

To compensate for this lack of manpower or budget, Leakmited relies on artificial intelligence. “It isolates the 20 to 30% of the network which contains 80% of the leaks, reducing the areas to be investigated. We can then monitor in a targeted manner with a single sensor which will be positioned in a relevant way based on the recommendations of the AI.” Hubert Baya Toda and his team of ten have set up a learning process for the technology.

The first step consists in integrating a hundred criteria aimed at describing the elements impacting a network and likely to cause a leak, whether it is the external environment and its evolution or the internal characteristics of the network. Then, to train the artificial intelligence, Leakmited set up a database.

“We started with 200,000 examples of leaks and currently have 600,000 over 300,000 kilometres, the equivalent of a third of the French network or as much as the Italian network.” From this reference data, the algorithm will generate a specific equation, extract a decision for each context that will be submitted to artificial intelligence, depending on the place, the environment, the season… “Sometimes we have missing or biased data. In this case, we have a replacement process to deliver a reliable prediction.”

Fundraising and prevention

Tested in France by Véolia, Suez or SAUR both in Redon (Ille-et-Vilaine) and in Rouen (Seine-Maritime) or Besançon (Doubs), the solution has also proven itself in Italy, Portugal and England. To deploy more widely, the start-up is currently raising one million euros to strengthen its sales team, projecting itself as well in France as in Latin America or in Africa where the waste of water is a higher level. .

Supported half by business angels, the fundraising is also based on a participatory platform “so that those who adhere to the protection of water resources can support us” insists Hubert Baya Toda. The funds will also finance new positions to consolidate the team in charge of the product. “We want to work from our customers’ feedback to improve the interface and make it more user-friendly.” These reinforcements will also work on new tools. A digital twin of the networks could thus become an essential player in the prevention of leaks.

Selected for you

Qubit Pharmaceuticals and Nvidia team up to accelerate drug discovery

We want to thank the writer of this article for this remarkable material

Leakmited relies on AI to protect water resources

Take a look at our social media profiles as well as other related pageshttps://yaroos.com/related-pages/