Artificial Intelligence for the maintenance and management of photovoltaic plants

The operation and maintenance (O&M) phase of a photovoltaic plant is receiving increasing attention from industry professionals. Indeed, it is a very long period, even more than 30 years, during which the plant operator has to meet complex commitments ranging, for example, from ensuring annual productivity that cannot fall below 2% compared to contracted levels, to resolution times for plant failures that, in the most serious cases, cannot exceed 72 hours.

Photovoltaic plants, especially utilities, can also extend over several hectares and be distributed across rather large territories, so that the maintenance phase can be quite challenging in terms of human resource commitment and warehouse management, with intuitive effects on the costs of the energy produced. Indeed, authoritative studies are beginning to show that OPEX costs are clearly increasing compared to the monotonous decrease in corresponding CAPEX costs. In this context, the ability to know the operational status of a plant at any given moment and then predict its evolution over time through the use of artificial intelligence techniques is becoming an increasingly inevitable operational necessity.

The first part of this workshop is dedicated to illustrating methods and techniques that can be applied to the O&M phase of photovoltaic plants: from energy forecasting techniques with their dual value in terms of productivity prediction and minimization of extreme event effects, to long-term prediction of plant failures with the possibility of organizing predictive rather than corrective maintenance procedures, to the use of control systems completely based on IoT logic. ENEA technicians who have developed the applied AI knowledge will discuss these topics. The second part of the workshop is dedicated to the illustration of an application case of the methods and techniques discussed in the first part. In the MARTA project - Advanced Monitoring and Management of Photovoltaic Plants in Network, funded by MIMIT through the Sustainable Growth Fund - Innovation Agreements, which is in its closing phase in these months, an Italian utility of about 5 MW has been revisited by applying artificial intelligence techniques also to protect the processes of delivering the produced energy to the power grid. Tea Tek spa, the plant operator, Molini Casillo, the plant owner, and ENEA, developer and provider of basic knowledge, will discuss this.

The third part aims to be a moment of discussion with the audience present, an informal Q&A session on the topics of AI, photovoltaic energy, and non-programmable energy sources.

Program

First part
Reliability and O&M of photovoltaic plants: economic and energy costs, and the role of Artificial Intelligence
Girolamo Di Francia, head of the Energy and Data Science Laboratory, ENEA

AI and robotics for energy forecasting and predictive maintenance on photovoltaic plants
Gabriele Piantadosi, ENEA

The world around: characterization of the plant's operating environment and cybersecurity issues
Saverio De Vito, ENEA

Second Part
Panel: Research and Applications for the reliable future of the energy system based on Photovoltaic and Non-programmable Sources

Introduction
Girolamo Di Francia, ENEA

Discussants:
Felice Granisso, Tea Tek and IGF CEO
Giovanni Campanile, Energy Manager Casillo Group
Giulia Monteleone, Director of the Renewable Energy Technologies Department ENEA
Elena De Luca, Coordinator of the Next Generation EU Sub-commission, Next Generation EU-PNIEC Technical Commission

Third Part
In conclusion, ENEA technicians, De Vito and Piantadosi, will be available along with Eng. Ippolito from Tea Tek spa to answer questions and delve deeper into the technologies that have been proposed.