Neural Network Based Procedure for Lightning Localization

Abstract:
This contribution proposes a new approach to locate lightning striking points by using the voltage waveform recorded by sensors placed on transmission lines. The method takes advantage of the use of Neural Networks (NN); the time domain voltage waveforms are transferred in the frequency domain with a Discrete Fourier Transform. Then, to reduce the dimensionality of the problem, it is performed a principal component analysis that gives a feature vector with only 10 components as input to the NN. Preliminary results, obtained in a simplified scenario, show an accuracy better than conventional location systems.