As of 2023, the global installed capacity of solar energy systems has reached 1.6 terawatts. This growth brings a critical question to the forefront:
How can millions of photovoltaic panels be monitored in a smarter, more reliable and more efficient way? The ZERODEFECT4PV project, which seeks to answer this very question, is one of the most significant R&D collaborations through which İnelso Energy engages with the international scientific community.
What is the ZERODEFECT4PV Project?
ZERODEFECT4PV is an international R&D project supported by the German Federal Ministry for Economic Affairs and Climate Action (BMWK) under the ERA-NET SES ENERDIGIT program. Conducted with support from TÜBİTAK in Turkey, the project includes Inelso Energy alongside partners Fraunhofer IFF (Germany) and BEIA Consult International (Romania).
The main objective of the project:
To develop fault detection and predictive maintenance systems for photovoltaic solar panels using AI algorithms.
What Kinds of Faults Occur in Solar Panels?
Research shows that the most common types of faults in photovoltaic systems can significantly reduce panel performance:
– Cell cracks: 0.9% to 60% power loss
– Potential Induced Degradation (PID): 25-60% power loss
– Soiling/Dust: 4.4-20% power loss
– Hotspots: 0.83-15.47% power loss
– Corrosion: 3-27% power loss
Early detection of these losses is critical for extending plant lifespan and minimizing production losses.
Inelso Energy’s Role in the Project
Inelso Energy manages the Turkish pilot site within ZERODEFECT4PV. Established in an operational solar farm in the Taurus Mountains, this pilot serves as a critical hub for the project's early hardware validation phase.
Data Collection Units (DCUs) developed by the Inelso team are mounted on the rear surface of photovoltaic panels to collect real-time module-level data on:
– Voltage measurement
– Current measurement
– Temperature tracking
It collects data in real time. This data is processed using artificial intelligence models at Fraunhofer IFF’s Integrated Operations Centre (IOC) and is used for fault detection and production forecasting.
AI Methods Utilized
Four key machine learning algorithms feature prominently in the project:
- Random Forest (RF): Exhibits superior performance in fault detection with a 99.4% accuracy rate. It can detect a wide range of faults such as partial shading, cell cracks, hotspots, and short circuits.
- Artificial Neural Networks (ANN): Reaching a 100% accuracy rate in simulations and real-world tests, ANN excels in modeling complex non-linear relationships.
- Support Vector Machine (SVM): An effective method for the classification of high-dimensional data.
- k-Nearest Neighbors (kNN): A simple yet effective algorithm for detecting short circuits, open circuits, and partial shading faults.
Three Countries, Three Pilot Sites
Pilot sites have been established in three different countries within the scope of the project:
🇹🇷 Turkey – Inelso Energy Pilot Site: Hardware validation is being conducted on polycrystalline panels at an operational solar power plant in the Taurus Mountains.
🇷🇴 Romania – BEIA Consult Pilot Site: A 6 kWp rooftop system monitored with SolarEdge inverters is utilized for testing AI models.
🇩🇪 Germany – Fraunhofer IFF Pilot Site: Two rooftop systems (10 kWp and 30 kWp) at the “Elbfabrik” complex are conducting the project's primary hardware and software tests.
The ZERODEFECT4PV project is taking significant steps in fault detection and predictive maintenance, one of the most critical challenges in the solar energy sector. Inelso Energy’s role in this international consortium is a tangible reflection of the company’s R&D-focused technology vision and its strong collaborations with global energy researchers.
We continue to work towards making solar energy systems smarter, more reliable, and more efficient.
Contact us to learn more about our SolarMetrica platform and SCADA solutions.
