Researchers Yue Zhao and Kang Pu from Stony Brook University—in collaboration with Ecosuite’s John Gorman and Philip Court, and leveraging historical datasets provided by Ecogy Energy—have devised a data-driven algorithm to detect physical anomalies in solar energy systems. This study aims to cut operations and maintenance (O&M) costs of solar projects by understanding and predicting long-term weather-related and inverter issues.
Through this effort, researchers trained anomaly detectors through a self-supervised learning approach trained on inverter and weather data. As such, with a holistic data-driven pipeline, the project not only employs a robust machine learning approach to model complex system behavior, but also ensures operational capacity across diverse data environments by opting for widely-available generation and weather data over non-standard measures.
The paper also takes an additional focus on long-term anomalies that evade the notice of many asset managers. The anomaly detectors, when applied to updating solar generation and weather data, will help accurately predict and diagnose underlying long-term physical issues weeks, or even years, before an asset manager can, if at all.
With such increased predictive knowledge of solar systems, these anomaly detectors can significantly reduce costs of O&M, a major component of project economics in solar development. There is great potential to optimize maintenance times and visits, as well as reduce the timespans of associated energy downtimes.
“Advanced warnings about DER assets directly from already paid-for edge-compute hardware just makes sense,” commented researcher John Gorman. “Adding these software superpowers immediately creates value but as part of a flexible ecosystem, our machine learning algorithms can also evolve. Being able to translate learnings from one system to the next is our next goal, unlocking value across a portfolio.”
Explore the Business Insights Journal for up-to-date strategies and industry breakthroughs!
