Detail publikace

Lightning Stroke Localization – A Time-Domain Approach Based on Evolutionary Optimization

KADLEC, P. MAREK, M. ŠTUMPF, M.

Originální název

Lightning Stroke Localization – A Time-Domain Approach Based on Evolutionary Optimization

Anglický název

Lightning Stroke Localization – A Time-Domain Approach Based on Evolutionary Optimization

Jazyk

en

Originální abstrakt

In the present paper we analyze a localization problem based on evolutionary optimization algorithms and an analytical EM-field-to-line coupling model. A computationally efficient, analytical model for calculating time-domain lightninginduced voltages serves as the forward solver in the optimization process. The model takes into account electric permittivity and conductivity of a lossy ground. The inverse problem is solved with the aid of a single-objective global optimization algorithm. It is demonstrated that the Covariant Matrix Adaptation - Evolution Strategy algorithm shows the best performance among considered state-of-the-art algorithms. The influence of other problem parameters, e.g. the size of the search space, the fluctuation of the return-stroke pulse width and amplitude, and the presence of noise, on the localization error is discussed. While the growing size of the search domain has a significant impact on the convergence properties of the optimization process, this is not the case for the signal distortion, whose influence can be virtually neglected.

Anglický abstrakt

In the present paper we analyze a localization problem based on evolutionary optimization algorithms and an analytical EM-field-to-line coupling model. A computationally efficient, analytical model for calculating time-domain lightninginduced voltages serves as the forward solver in the optimization process. The model takes into account electric permittivity and conductivity of a lossy ground. The inverse problem is solved with the aid of a single-objective global optimization algorithm. It is demonstrated that the Covariant Matrix Adaptation - Evolution Strategy algorithm shows the best performance among considered state-of-the-art algorithms. The influence of other problem parameters, e.g. the size of the search space, the fluctuation of the return-stroke pulse width and amplitude, and the presence of noise, on the localization error is discussed. While the growing size of the search domain has a significant impact on the convergence properties of the optimization process, this is not the case for the signal distortion, whose influence can be virtually neglected.

Dokumenty

BibTex


@article{BUT161030,
  author="Petr {Kadlec} and Martin {Marek} and Martin {Štumpf}",
  title="Lightning Stroke Localization – A Time-Domain Approach Based on Evolutionary Optimization",
  annote="In the present paper we analyze a localization
problem based on evolutionary optimization algorithms and an analytical EM-field-to-line coupling model. A computationally efficient, analytical model for calculating time-domain lightninginduced voltages serves as the forward solver in the optimization process. The model takes into account electric permittivity and conductivity of a lossy ground. The inverse problem is solved with the aid of a single-objective global optimization algorithm. It is demonstrated that the Covariant Matrix Adaptation - Evolution Strategy algorithm shows the best performance among considered state-of-the-art algorithms. The influence of other problem parameters, e.g. the size of the search space, the fluctuation of the return-stroke pulse width and amplitude, and the presence of noise, on the localization error is discussed. While the growing size of the search domain has a significant impact on the convergence properties of the optimization process, this is not the case for the signal distortion, whose influence can be virtually neglected.",
  chapter="161030",
  doi="10.1109/TEMC.2020.2964059",
  howpublished="print",
  number="5",
  volume="62",
  year="2020",
  month="january",
  pages="2111--21188",
  type="journal article in Web of Science"
}