Forecasting cross-border power transit: a bachelor thesis developed with ČEPS

Energy systems offer student projects a demanding combination of physical processes, changing conditions and large public datasets. I supervised Martin Horešovský's bachelor thesis on forecasting cross-border electricity transit, developed with external consultant Petr Souček from ČEPS. Martin defended the thesis with the highest grade on 16 June 2026, after taking first place in the bachelor section of the Faculty of Applied Sciences Student Conference.

The thesis, Forecasting Cross-Border Power Transit from Open Data Sources, combined time-series modelling with open energy and weather data. Martin designed an automated pipeline for aligning, cleaning and transforming the inputs, compared prediction models and evaluated their use for forecasting electricity transit across the borders of the Czech Republic.

The work was tied to professional practice from the outset. I supervised the academic and machine-learning side, while Petr Souček contributed the perspective of ČEPS, the Czech transmission system operator. This gave the project a concrete engineering question and a setting in which data quality, prediction horizons and changing energy sources matter as much as model selection.

Before the defence, Martin presented the work at the 2026 Student Conference of the Faculty of Applied Sciences, where he placed first in the bachelor section and received a GK Software sponsor award.

The complete thesis PDF is available directly from the university repository. The related STAG record includes the reviews and defence record; Martin successfully defended the work on 16 June 2026 with the grade Excellent.

First page of Martin Horešovský's bachelor thesis Forecasting Cross-Border Power Transit from Open Data Sources — First page of the bachelor thesis published in the University of West Bohemia's thesis repository.

Abstract

The dynamic expansion of renewable energy sources in Europe introduces increasing volatility into transmission grids, leading to frequent and massive unplanned electrical energy transit flows across the Czech Republic. This bachelor thesis focuses on the design and implementation of a computational system for predicting these physical flows using machine learning methods and open data sources. For modeling purposes, data from the ENTSO-E Transparency platform and the Open-Meteo API were automatically acquired and harmonized. The problem was framed as a time series regression task. During the experiments, several regression models (e.g., ElasticNet, Lasso, SGDRegressor) were evaluated on various feature sets for all four cross-border profiles of the Czech Republic. The results demonstrate that machine learning models can highly accurately approximate the behavior of the European grid solely based on historical data, even without knowledge of the internal physical topology. The best performance was achieved in predicting the total transit balance of the Czech Republic, reaching a coefficient of determination R² = 0.9589. Furthermore, the proposed system was successfully validated during a multi-week deployment in a real-time operational environment. The thesis proves that a data-driven approach represents an effective tool for short-term prediction of grid load, which can enhance grid security and the overall economic efficiency of dispatch management.

Abstract

Links