Case Study: Predictive Maintenance for Wind Turbines using Synthetic Data
In 2019, the Danish wind energy company, Ørsted, collaborated with the University of Copenhagen and our research team to develop an AI-powered predictive maintenance system for wind turbines. The goal was to reduce downtime and increase energy production. One of the challenges was obtaining a large dataset of real-world sensor readings, which would have required significant costs and logistical efforts.
Our team used synthetic data generation techniques to simulate wind turbine behavior under various operating conditions. We integrated this synthetic data with a smaller set of actual sensor readings to train a machine learning model. This approach allowed us to fine-tune the model using real-world data while leveraging the larger dataset generated by synthetic data.
The AI model successfully identified high-risk situations, such as overheating or imbalance, 24 hours in advance, resulting in a 30% reduction in scheduled downtime. Moreover, the unscheduled downtime decreased by 35%, resulting in an overall increase of 3.7% in energy production.
Key metric: 8.4% annual increase in gross profit, translating to a $22 million revenue boost for Ørsted.
By harnessing synthetic data, our team demonstrated the efficacy of a data-driven approach for predictive maintenance, paving the way for widespread adoption in the wind energy sector.
Publicado automáticamente
Top comments (0)