Prediction, optimisation, and energy systems.

Latest publication

Day-Ahead Heat Demand Prediction Using BiLSTM-Based Hybrid Neural Networks

Citation: J. Ullmann, J. Závorka, J. Horák, J. Jiřičková, and R. Škoda, “Day-Ahead Heat Demand Prediction Using BiLSTM-Based Hybrid Neural Networks,” in Proceedings of the 26th International Scientific Conference on Electric Power Engineering (EPE 2026), Opole, Poland, 18-20 May 2026.

DOI: 10.1109/EPE68416.2026. IEEE Xplore Document ID: 11537536.

This is the strongest AI-first publication signal for the portfolio: a forecasting-oriented contribution built around BiLSTM-based hybrid neural networks and day-ahead heat-demand prediction.

Poster / presentation

Data-Driven Methodology and Tools for Neural-Network-Based Operational Optimization of Small Modular Reactors at the Texas A&M RELLIS Campus

Status: Published. Type: Lecture/poster abroad.

Authors: Jan Ullmann, Jiří Závorka, Pavel V. Tsvetkov, Radek Škoda.

Record: ID 43949778 | RIV-Ne | Year: 2026. Record owner: Jitka Machová (FEL).

This item supports the AI and optimisation direction of the site: neural-network-based tools for operational optimisation in an applied campus energy-system context.

Journal article

Flexible nuclear district heating with the Teplator concept: ORC cogeneration, load shifting, and data center integration

Authors: Jan Ullmann, David Mašata, Jiří Závorka, Jana Jiřičková, Radek Škoda

Journal: Sustainable Energy Technologies and Assessments, Volume 87, March 2026, 104912

DOI: 10.1016/j.seta.2026.104912

Decarbonizing the heat sector is a major energy-transition challenge. This study evaluates district heating with ORC cogeneration, thermal load shifting, and data-center integration using real Pilsen district-heating data.