Future electricity grids created with federated machine learning
This project addresses the critical need for improved tools for energy companies in analyzing their customers behaviour. By developing a federated machine learning model for analyzing electricity grid data the project will create new insights for the energy companies.

Image (AI-generated): A diverse Swedish neighborhood featuring a mix of housing types, illustrating varied patterns of electricity usage.
The project aims to create better, data-driven decision support for developing the electricity system, enabling sufficient electrical power to be available in the right place, at the right time, and in the right amount.
A key component in electricity grid scenarios and forecasts is the ability to understand how customers' electricity demand may change. This is crucial for planning electricity infrastructure development.
Currently, many electricity network companies, especially smaller ones, face challenges in creating effective scenarios and forecasts due to limited data. Since the data is often protected by regulations, sharing it is difficult or even impossible. However, by utilising federated learning, electricity network companies can collaboratively train machine learning models on their combined data without ever sharing sensitive information.
This project aims to develop such federated machine learning models, enabling better analysis for more optimal management of electricity network operations and development.
Challenges
- The increasing demand for electricity due to industrial expansion and electrification.
- Difficulties in granting requests for increased electrical power.
- The challenge of creating accurate scenarios and forecasts for future power needs.
- Restrictions on sharing electricity network data due to security and confidentiality.
Project purpose
The project's purpose is to improve electricity network companies' ability to analyze and optimize energy flows by developing a machine learning model for time series data from electricity networks using federated machine learning. This technology enables training a model on data from various parties without sharing the data itself, addressing data privacy concerns.
The project will lay the foundation for a national de facto standard for using federated machine learning to create future load profiles, considering various customer metadata.
Expected outcomes
- A proposal for a standardized data structure for electricity network data
- A proposal for a federated ML platform for electricity network companies
- A prototype of an ML model for detecting electricity customers and behaviours
- Standardized load profiles for different electricity customers (initially for a limited number)
What is Federated machine learning?
Federated machine learning is a technique where multiple stakeholders collaboratively train a machine learning model without sharing their raw data. Instead, each device or silo trains the model locally and only shares updates (like gradients), keeping the data private and secure.
Facts
Funding: Vinnova
Participants:
- AI Sweden
- Herrljunga Elektriska
- Jönköping Energi
- Kinnekulle Energi
- Lerum Energi
- Lokalkraft Sverige
- Mölndal Energi
- Öresundskraft
Project period: 2024-10-01 - 2025-10-31
For more information, contact

Related projects

Simulation and visualization of Sweden's future electricity system

Federated machine learning for creating power load profiles

E-Charge 2: Pioneering AI solutions for heavy transport electrification
