A software tool developed by Stony Brook University uses self-supervised learning to detect long-term damage to solar power equipment weeks or even years before it is discovered by manual inspection.
Stony Brook University researchers, in collaboration with Ecosuite and Ecogy Energy, have developed a self-supervised machine learning algorithm designed to identify physical anomalies in solar energy systems. The project aims to reduce operations and maintenance (O&M) costs, which remain a major hurdle to project economics as the industry expands.
Researchers Yue Zhao and Kang Pu trained an anomaly detector using a comprehensive pipeline that leverages historical datasets provided by Ecogy Energy and integrates inverter performance and weather data. This approach avoids non-standard measures and instead chooses widely available power generation and environmental data to ensure that tools continue to operate reliably across diverse data environments.
This research focuses specifically on long-term anomalies involving underlying physical issues that often evade asset manager notification until significant downtime or hardware failure occurs. When applied to solar power generation and weather data updates, this detector is designed to predict and diagnose these problems weeks or even years in advance.
“More accurate and timely recognition will significantly improve the efficiency and effectiveness of O&M practices,” the researchers said.
The study outlines three main areas of impact:
-
Efficient scheduling: Optimize maintenance site visits to address specific issues detected and reduce unnecessary truck trips.
-
Asset Lifespan: On-time hardware maintenance significantly extends the lifespan of equipment and reduces the need for expensive and premature replacements.
-
Protect your bottom line: Addressing system issues before failures can minimize losses in energy production.
As your solar portfolio grows, the ability to automate the detection of performance deficiencies will become important. According to recent industry data, U.S. solar installations lost an average of $5,720 per MW due to equipment issues in 2024 alone. These “software superpowers,” as researcher John Gorman calls them, provide advanced alerts directly from the edge computing hardware you've already paid for.
“Adding the super-powerful capabilities of these software will create immediate value, but as part of a flexible ecosystem, our machine learning algorithms can also evolve,” Gorman said. “Being able to translate what we learn from one system to the next is our next goal, allowing us to unlock value across our portfolio.”
Researchers believe that advances in these algorithms will enable asset managers to move from reactive maintenance to proactive and predictive models that maximize the financial and operational health of renewable energy assets.
This content is copyrighted and may not be reused. If you would like to collaborate with us and reuse some of our content, please contact us at editors@pv-magazine.com.
Popular content
