On-Line Frequency Forecasting using Convolutional Neural Networks |
|---|
| Théo Chacou Bertoldi, Viktoriya Mostova, Silvia Iuliano, Alfredo Vaccaro |
- Abstract:
- In modern power systems the replacement of large synchronous generators with distributed inverter-based resources is lowering the power system inertia, making electrical grids more vulnerable to dynamic perturbations. For this purpose the European Network of Transmission System Operators for Electricity promoted the enhancement of the grid operation tools with specific functions for on-line estimation of the power system inertia, which are extremely useful in detecting critical thresholds and triggering proper mitigation strategies. For this purpose, the development of reliable frequency predictive models represents an essential requirement for enabling system inertia estimation. To try and address this issue, this paper explores the role of Convolutional Neural Networks (CNN) for on-line frequency estimation from grid measurements. The main idea is to model the grid frequency deviations by a CNN-based identification technique, which allows inferring the main parameters ruling the power system dynamics. Detailed simulation results obtained on several case studies demonstrate that the CNN model is able to detect data patterns, and discover hidden relationships maintaining low estimation errors even for multi-step ahead frequency predictions, thus offering a valuable tool for inertia estimation in modern power systems, especially with the increasing share of renewable energy.
- Download:
- IMEKO-TC6-2025-044.pdf
- DOI:
- 10.21014/tc6-2025.044
- Event details
- IMEKO TC:
- TC6
- Event name:
- TC6 M4Dconf2025
- Title:
2025 IMEKO TC-6 International Conference on Metrology and Digital Transformation
- Place:
- Benevento, ITALY
- Time:
- 03 September 2025 - 05 September 2025