Towards a Digital Twin Power System: A Case Study for Wind Power Forecasting

This paper presents a significant step towards realizing a digital twin power system, focusing on a case study for wind power forecasting. We propose a novel framework that integrates digital twin technology with a Gated Recurrent Unit (GRU) neural network, termed DT-GRU. This approach leverages real-time data processing capabilities and adaptive modeling to enhance wind power forecasting accuracy. Based on the digital twin based forecasting framework, the online learning mechanism is tailored to update the DT-GRU model in real time. Using real-world high-resolution data, we evaluate the DT-GRU model against traditional machine learning approaches for one-hour and two-hour ahead forecasts. Results demonstrate that our digital twin-based model consistently outperforms conventional methods across multiple error metrics, particularly in capturing rapid fluctuations in wind power output.