This is the authors' abstract. We don't add key points for this paper.
The paper presents a new approach for lightning location and peak current estimation based on Deep Learning (DL) algorithms. The basic idea is to use the time domain waveforms of the overvoltages induced by lightning strikes on transmission or distribution lines as features to regress both the location and the channel base current peak of a lightning event. Starting from previous research findings, the performance of the method have been further increased by proposing three different DL models, working either in the frequency or in the time domain. All models are trained with data simulated with the Lightning Power Electromagnetic Simulator for Transient Overvoltages (LIGHT-PESTO). The performances of the methods are first assessed on the same test case as the one presented in a previously published paper to show their increased effectiveness. Then, the best performing model among the three is tested on a more realistic network to show that, even in presence of much more complex systems, its performances are not reduced. • Deep Learning models developed to estimate lightning location and peak current. • Models training with simulated voltage waveforms induced on overhead power lines. • Realistic three-phase distribution system with surge arresters above a lossy ground. • Improved accuracy compared to previous algorithms with the same purposes. • Competitive performances with conventional Lightning Location Systems.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Dodge et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: