Leveraging federated learning for renewable scenario generation

What is it about?

Fed-LSGAN, a federated deep generative learning framework, is proposed for renewable scenario generation. Here, federated learning enables the proposal to generate scenarios in a privacy-preserving manner by transferring model parameters, rather than all data.

Featured Image

Photo by RawFilm on Unsplash

Photo by RawFilm on Unsplash

Why is it important?

1) This paper proposes Fed-LSGAN, a novel federated learning-based framework for renewable scenario generation. To the authors’ knowledge, this is the first work to leverage federated learning for scenario generation. Here, federated learning learns a shared global model in a central server from renewable power plants at network edges, which enables the Fed-LSGAN to generate scenarios in a privacy-preserving manner without sacrificing the generation quality by transferring model parameters, rather than all data. 2) The Fed-LSGAN features on a new design of the LSGANs-based deep generative model in a distributed fashion for generating scenarios that conform to the distribution of historical data through fully capturing the temporal and spatial characteristics of renewable powers, which leverages the least squares loss function to improve the training stability and the quality of generated scenarios. 3) Extensive tests have been performed using a real-world dataset to examine the effectiveness of the Fed-LSGAN, which demonstrates that Fed-LSGAN is superior to state-of-the-art centralized methods. Besides, an experiment with different control parameters in federated settings is designed to examine the robustness of our method.