A Divide-and-Conquer approach for denoising and modeling the CN Tower lightning current derivative signal

Abstract

The CN Tower is a transmission hub and an instrumented tower for the measurement of the lightning return stroke current derivative. The recorded data are corrupted by different kinds of noise, and need to be denoised for accurate determination of the lightning return-stroke current waveform parameters. A new Divide-and-Conquer denoising approach that imitates the Basis Pursuit method and the Newton-Raphson technique has been developed. This paper describes the new process of denoising the recorded signals. First, the current derivative is preprocessed to eliminate the noise outside the lightning return-stroke active region and reduce the presence of the high frequencies inside the active region. Then, by marching on both the graphs of the current derivative and its integral, the noise due to reflections is localized and removed. By this process the SNR improved by 35 dB and the lightning current and current derivative parameters are calculated automatically with a high precision. Furthermore, using the calculated parameters the data is curve fitted to Heidler function, which results in a model for the measured lightning current derivative with an infinite SNR