A novel method for ground-based VLF/LF single-site lightning location. (15th June 2022)

• Record Type: Journal Article
• Authors: Wang, Jiaquan; Xiao, Fang; Yuan, Shangbo; Song, Jiajun; Ma, Qiming; Zhou, Xiao
• Is Part Of: Measurement. Volume 196(2022)
• Publication Date: 2022-06-15
• Subjects: Single-site -- Lightning location -- Lightning waveform -- Deep learning -- Regression model; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2022.111208
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A novel lightning geolocation method for single-site is proposed. The SLL-ResNet including three feature extraction branches and a feature fusion network is employed to locate lightning. The model achieves a prediction error of 2.75–13.08% for the propagation distance of LEMP within a range of 1000 km. The SLL-ResNet is available for all types of LEMP signals. Abstract: A single-site lightning electromagnetic pulse (LEMP) localization approach based on deep learning was proposed and practiced. The approach was based on a large amount of ground-based lightning location and waveform data in the VLF/LF frequency band. A model for predicting the propagation distance of LEMP based on a deep learning method was proposed. The model used multiple types of LEMP waveform data as well as location data for parameter learning. The new model for detecting lightning activity has been validated against ADTD systems. The verification results for thunderstorms in the range of 1000 km show that the relative error of the model for the prediction of signal propagation distance was 2.75–13.08%. Affected by the azimuth calculation error, the relative error of single-site geolocation was 4.91–15.26%.