Optimizing Federated Learning for Scalable Power-demand Forecasting in Microgrids

2025-08-11

AI Summary

To address non-IID data bias and scalability limitations with thousands of clients in federated learning for microgrid and city-scale electricity demand forecasting, this paper proposes a lightweight federated optimization framework based on exponentially weighted loss. The method trains deep neural networks locally at edge devices and employs dynamic weighted aggregation to mitigate model bias induced by non-IID data, significantly reducing communication and computational overhead. Evaluated on the real-world OpenEIA dataset, it achieves 12.6–19.3% higher prediction accuracy than ARIMA and single-user DNN baselines, with 41% lower training time and support for over 1,000 heterogeneous clients. Experimental deployment on Raspberry Pi edge clusters and a pseudo-distributed environment demonstrates strong privacy preservation, high prediction accuracy, and exceptional scalability. This work establishes a practical, scalable paradigm for federated modeling in large-scale energy systems.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Real-time monitoring of power consumption in cities and micro-grids through the Internet of Things (IoT) can help forecast future demand and optimize grid operations. But moving all consumer-level usage data to the cloud for predictions and analysis at fine time scales can expose activity patterns. Federated Learning~(FL) is a privacy-sensitive collaborative DNN training approach that retains data on edge devices, trains the models on private data locally, and aggregates the local models in the cloud. But key challenges exist: (i) clients can have non-independently identically distributed~(non-IID) data, and (ii) the learning should be computationally cheap while scaling to 1000s of (unseen) clients. In this paper, we develop and evaluate several optimizations to FL training across edge and cloud for time-series demand forecasting in micro-grids and city-scale utilities using DNNs to achieve a high prediction accuracy while minimizing the training cost. We showcase the benefit of using exponentially weighted loss while training and show that it further improves the prediction of the final model. Finally, we evaluate these strategies by validating over 1000s of clients for three states in the US from the OpenEIA corpus, and performing FL both in a pseudo-distributed setting and a Pi edge cluster. The results highlight the benefits of the proposed methods over baselines like ARIMA and DNNs trained for individual consumers, which are not scalable.

Problem

• Optimizing federated learning for scalable power-demand forecasting
• Addressing non-IID data and computational efficiency in FL
• Improving prediction accuracy while minimizing training costs

Innovation

• Federated Learning for privacy-sensitive DNN training
• Exponentially weighted loss improves prediction accuracy
• Scalable optimization for non-IID data across edge-cloud