Player of the week: Heliocity puts AI at the service of the performance of solar power plants – PV Magazine

Founded in September 2020, the Grenoble-based “deeptech” start-up Heliocity has developed HelioFlash, a base of artificial intelligence algorithms that makes it possible to diagnose the proper functioning of a solar power plant on buildings while providing recommendations to improve the installation.

December 11, 2023 Francois Puthod

Heliocity is a spin-off of INES (National Institute of Solar Energy) and offers solutions to unlock and support the potential of solar energy in built environments thanks to innovative algorithms that enable remote diagnostic analysis.

As the adventure begins, a paradoxical observation stands out: harnessing solar energy in a built environment presents real difficulties, even though most of the planet’s energy needs are concentrated in urbanized areas. “For ten years we have been looking for ways to bring solar systems and buildings together,” recalls Émeric Eyraud, CEO of Heliocity, in an interview with pv magazine France. The structural environment places high demands on a solar power plant, particularly with regard to the thermal interaction with the building. If we simplify: two degrees more temperature of a module, that’s 1% less production.”

Player of the week Heliocity puts AI at the serviceScreenshot of Heliocity’s Software as a Service (SaaS) platform, a cloud-based service accessible via an app.

Image: Heliocity

Knowing the temperature of a module is crucial. It is a first-order factor in knowing what it can produce, just like solar radiation, which is also little known due to the lack of radiation probes in power plants on buildings. “In fact, you have a solar power plant, you don’t know how much sun it sees, you don’t know the temperature and you have to pre-judge what it should produce…” summarizes the CEO.

Patented algorithms

Heliocity has “cracked” this difficult problem, which does not escape large ground-based power plants despite the presence of sensors, by developing its HelioFlash solution. The latter is based on innovative, patent-protected algorithms that combine multi-physical and multi-scale modeling of buildings, solar systems and the environment with complex data analysis methods adapted to monitoring data. To date, it has been used by around thirty operators and managers of solar parks, including the majority of the ten largest operators in France.

1702289306 759 Player of the week Heliocity puts AI at the serviceCleaning process as part of a qualification test of the dirt detection algorithms.

Image: Heliocity

“Of course we will try to understand what exactly the solar power plant sees in relation to the environment, knowing that there is no probe, and what its electrical response is, that is, what it produces,” explains Émeric Eyraud. By comparing what it should have produced with what it can actually produce, the algorithms can then highlight weaknesses in performance with unmatched detection accuracy, but most importantly, determine the cause: a shading problem, dirt in a cell, incorrect layout, etc. Example and can help with recommendations.

“Digital Twin”

Monitoring solutions are not enough to offer such a service. The innovation demonstrated by Heliocity lies in the ability to carry out a complete diagnosis of a solar power plant within 48 hours, based on existing data automatically provided via ad hoc gateways. “We are like a little extra brain that gives the operators or owners of a power plant an understanding of the plant’s performance. And above all, tell them what they need to do to make it work better: when to clean the panels or when to replace them if, for example, they do not meet the manufacturer’s guarantees, explains Émeric Eyraud. To my knowledge, we are the only ones who can say with tracking data alone: ​​”Here you have shadows, there you have dirt, there you have a problem of this kind, with this module, at this point”.

1702289309 206 Player of the week Heliocity puts AI at the serviceDevelopment of the performance of a photovoltaic power plant.

Image: Heliocity

Data collection is completely remote, with no visit to the facility required. Two types of information are necessary for their algorithmic use. Descriptive information on around fifteen variables (location and performance of the system, configuration, orientation and tilt of the modules, module types, inverter model, etc.) used to create a “digital twin”. These fixed data are supplemented by monitoring data (voltage, current) and environmental data (temperature, wind, irradiation probes). The solutions offered by Heliocity are aimed at solar power plants with an output of more than 35 kWp and can be used ad hoc – for a few hundred euros depending on the size of the system – or as a package with a subscription. What is also innovative is that the audit is carried out completely online, thanks to access codes to a platform on which the customer can view the result of the diagnostic analysis in detail.

Fundraising of 1.2 million euros in 2024

As technology continues to improve, Heliocity as a map road has set two priorities: fusing available data to enable ever more accurate diagnosis of the asset and automating the interpretation of recommendations. “We are also working on a “field” application. It could make the work of caregivers easier while validating the anomalies detected by our algorithms,” adds Émeric Eyraud.

Meanwhile, the winning company of the i-Lab 2023 competition hopes to raise €1.2 million in funding in 2024, primarily to accelerate its commercial development. “We have just started, we are still at a young beginning,” recognizes the CEO of Heliocity. Our team, which mainly consists of technicians and engineers, consists of 7 people. Now that we have developed and industrialized the solution, we need to approach the market, especially the German one, which is at least seven times larger than the French one.”

With the new European legislation requiring the installation of solar panels on the roofs of commercial and public buildings by 2025 and on the roofs of residential buildings by 2029, the sales target of five million euros that the company has set for itself will be reached in the next Five years should be fulfilled.

This content is copyrighted and you may not reuse it without permission. If you would like to collaborate with us and reuse our content, please contact our editorial team at the following address: [email protected].