French startup offers AI algorithm package for rooftop PV monitoring


From pv magazine France

Heliocity, a spinoff of France's National Solar Energy Institute (INES), has developed a series of algorithms to remotely conduct diagnostic assessments of rooftop PV systems.

“For 10 years, we looked for ways to bring together solar installations and buildings,” Heliocity CEO Émeric Eyraud told pv magazine France. “The building environment is quite demanding for PV systems, particularly in terms of its thermal interaction with the building itself. If we simplify, two more degrees of temperature in a module means 1% less production.”

Heliocity claims to have “cracked” this issue – which can also affect large ground-based power plants, despite the use of sensors – with its HelioFlash solution. The patented algorithms combine multi-physics and multi-scale modeling of buildings, solar systems, and the environment with complex data analysis methods adapted to data monitoring. The company said that approximately 30 solar park operators, including most of France's top 10, have already deployed it.

The algorithms compare expected PV system output with actual production, detecting and pinpointing underperformance issues such as shading, dirt, and incorrect layout, while offering recommendations for remediation.

Popular content

Heliocity's solution conducts comprehensive PV system assessments within 48 hours, leveraging existing data delivered through automated ad hoc gateways.

“We are like a little extra brain, which will give the operators or owners of PV systems an understanding of the performance of the installation. And above all, tell them what to do to make it work better, for example, when to clean the panels or when to replace them if they do not perform in accordance with the manufacturer's guarantees,” Eyraud said. “To my knowledge, we are the only ones to date who can say, just with tracking data. Here you have shading, there you have dirt, there you have a problem of this nature, with this module, in this place.”

Remote data collection eliminates the need for on-site visits, but two details are needed for algorithmic exploitation. Descriptive information on approximately 15 variables is used to create a “digital twin,” complemented by monitoring data, including voltage, current, temperature, wind, and irradiation probes.

The solution is aimed at PV systems above 35 kW in size and can be used on an ad hoc basis for a few hundred euros, depending on the size of the installation, or as a subscription package. Clients access detailed diagnostic analysis results through an online platform using access codes.

This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: