Jonas Sievers

I am a Ph.D. student at the Karlsruhe Institut of Technology (KIT) affiliated with the Institute for Data Processing and Electronics (IPE), with a primary focus on machine learning applications in energy systems. Throughout my research I explore different machine learning techniques, including federated learning, reinforcement learning, and time series forecasting, as they are increasingly important in the optimization of energy management systems.

Federated learning enables collaborative model training without sharing sensitive data, allowing various energy devices to work together enhancing data security, privacy, latency, and bandwidth efficiency within the underlying communication network.

Reinforcement learning enables energy management systems to make dynamic decisions through learning from interactions with the environment, leading to adaptive control strategies that optimize energy consumption, emissions, and operational costs.

Time series forecasting, driven by machine learning algorithms, utilizes historical energy data to make accurate predictions, helping energy managers plan and allocate resources more efficiently, while improving grid stability.

news

Oct 14, 2023	I gave a talk at “Energy Status Data” titled “Connected Energy Management Systems to supply smart production sites with renewable energies”.
Sep 13, 2023	Our paper “Secure short-term load forecasting for smart grids with transformer-based federated learning” got accepted at ICCEP 2023.
Feb 8, 2023	Our paper “A Systematic Literature Review on Data-Driven Residential and Industrial Energy Management Systems” got accepted at Energies 2023.

selected publications

Energies
A Systematic Literature Review on Data-Driven Residential and Industrial Energy Management Systems

J. Sievers, and T. Blank

Energies 2023, Feb 2023

Abs Bib PDF

The energy transition and the resulting expansion of renewable energy resources increasingly pose a challenge to the energy system due to their volatile and intermittent nature. In this context, energy management systems are central as they coordinate energy flows and optimize them toward economic, technical, ecological, and social objectives. While numerous scientific publications study the infrastructure, optimization, and implementation of residential energy management systems, only little research exists on industrial energy management systems. However, results are not easily transferable due to differences in complexity, dependency, and load curves. Therefore, we present a systematic literature review on state-of-the-art research for residential and industrial energy management systems to identify trends, challenges, and future research directions. More specifically, we analyze the energy system infrastructure, discuss data-driven monitoring and analysis, and review the decision-making process considering different objectives, scheduling algorithms, and implementations. Thus, based on our insights, we provide numerous recommendations for future research in residential and industrial energy management systems.
@article{Energies.2023, title = { A Systematic Literature Review on Data-Driven Residential and Industrial Energy Management Systems}, author = {Sievers, J. and Blank, T.}, journal = {Energies 2023}, volume = {16}, issue = {4}, pages = { 1688}, numpages = {0}, year = {2023}, month = feb, publisher = ieeexplore, doi = {https://doi.org/10.3390/en16041688}, dimensions = {true}, }
ICCEP
Secure short-term load forecasting for smart grids with transformer-based federated learning

J. Sievers, and T. Blank

International Conference on Clean Electrical Power (ICCEP), Sep 2023

Abs Bib PDF Code

Electricity load forecasting is an essential task within smart grids to assist demand and supply balance. While advanced deep learning models require large amounts of high-resolution data for accurate short-term load predictions, fine-grained load profiles can expose users’ electricity consumption behaviors, which raises privacy and security concerns. One solution to improve data privacy is federated learning, where models are trained locally on private data, and only the trained model parameters are merged and updated on a global server. Therefore, this paper presents a novel transformer-based deep learning approach with federated learning for short-term electricity load prediction. To evaluate our results, we benchmark our federated learning architecture against central and local learning and compare the performance of our model to long short-term memory models and convolutional neural networks. Our simulations are based on a dataset from a German university campus and show that transformer-based forecasting is a promising alternative to state-of-the-art models within federated learning.
@article{ICCEP.2023, title = {Secure short-term load forecasting for smart grids with transformer-based federated learning}, author = {Sievers, J. and Blank, T.}, journal = {International Conference on Clean Electrical Power (ICCEP)}, volume = {}, issue = {}, pages = { 229-236}, numpages = {0}, year = {2023}, month = sep, publisher = ieeexplore, doi = {https://doi.org/10.1109/ICCEP57914.2023.10247363}, dimensions = {true}, }