Towards intelligent operation and maintenance in large-scale sustainable energy systems

Yingying Zhao (Fudan University )

Friday 30th September, 2022 15:00-16:00 Zoom

Abstract

Sustainable energy, as an alternative to fuel energy, has readily received concerns from society because of its merits of zero carbon emissions. Owing to the availability of sensor data, operation and maintenance (O&M) of sustainable energy systems now becomes more intelligent. In particular, data-driven anomaly detection approaches have gained growing interest. However, effectively and efficiently detecting anomalies is not a simple task. The deficiency of labeled data poses a major challenge in learning an accurate anomaly detector. Moreover, the large volume of video data collected from numerous sub-system components incurs a high computational cost and thus making it difficult to perform real-time system anomaly detection.

This talk will focus on investigating and developing: 1) unsupervised learning-based approaches to achieve accurate anomaly detection, which can improve the operation reliability of large-scale sustainable systems; 2) a reinforcement learning-based energy-efficient framework to reduce the O&M video data consumption; consequently, we can achieve more intelligent O&M. The proposed methods have been deployed in several large-scale solar and wind farms in China and support day-by-day system operation.

Add to your calendar

Download event information as iCalendar file (only this event)