Decentralised Secure Modelling of Electrolyzers with Federated and Transfer Learning
conference paper
The rapid evolution of electrolyzer technologies demands efficient and secure methods for exchanging sensitive information among distributed stakeholders. In this paper, we propose a novel solution for secure exchange of information between electrolyzer Operators while preserving data privacy. Our approach leverages Federated and Transfer Learning techniques, incorporating local predictive models and a central global model as a two-tier learning system. By training local models on individual electrolyzer system, we enable the central global model to aggregate the learned features without directly accessing the sensitive data. This approach not only alleviates issues generated by anti-competitiveness regulations, as many businesses don’t want to share data, but also enables efficient information exchange among electrolyzer operators in a distributed context. We demonstrate the effectiveness of our proposed methodology through comprehensive experiments, showcasing its potential to model and optimize electrolyzer performance without compromising data confidentiality. Our solution has broad implications for the development and deployment of electrolyzer systems, providing a secure and efficient means for stakeholders to collaborate and advance sustainable hydrogen production technologies.
TNO Identifier
1005355
Publisher
IEEE
Source title
IEEE PES ISGT Europe 2024 conference
Files
To receive the publication files, please send an e-mail request to TNO Repository.