Federated Learning-Based Energy Forecasting and Trading Platform for Decentralized Renewable Energy Markets
Ramakrishna S S Nuvvula, Polamarasetty P Kumar, Praveena Akki, Syed Riyaz Ahammed, Sudheer Reddy. J, R Hushein, Ahmed Ali
Abstract
Decentralized renewable energy markets are witnessing rapid growth, driven by the increasing adoption of renewable energy sources and the need for sustainable energy solutions. In this paper, we propose a federated learning-based energy forecasting and trading platform tailored for decentralized renewable energy markets. Leveraging the federated learning framework, our platform enables accurate and privacy-preserving energy forecasting while facilitating efficient energy trading and grid management. Through empirical evaluations using real-world data, we demonstrate the superior performance of our federated learning model in predicting energy generation levels compared to traditional baseline models. Our model achieves a Mean Absolute Error (MAE) of 12.36 MW and a Root Mean Squared Error (RMSE) of 16.82 MW, outperforming Autoregressive Integrated Moving Average (ARIMA) and Prophet models. Furthermore, our platform exhibits scalability and robustness, capable of handling diverse and distributed datasets while maintaining performance in the face of data distribution shifts and individual model failures. With an average training time of 6 hours per round and a total of 20 communication rounds, our platform efficiently utilizes distributed computing resources while preserving data privacy. Additionally, our platform enables more efficient energy trading by providing accurate forecasts of energy generation levels, leading to potential cost savings of up to 15% for market participants. Moreover, it enhances grid management and stability by reducing grid congestion by up to 20% and improving resource utilization efficiency by up to 25%. Overall, our federated learning-based energy forecasting and trading platform offers a transformative solution for decentralized renewable energy markets, promoting efficiency, reliability, and sustainability. Its adoption has the potential to revolutionize energy markets and accelerate the transition towards a clean and renewable energy future.