Key points are not available for this paper at this time.
A method for detecting series arcs from time-frequency analysis of the current derivative is proposed. Mathematical morphology is used to select arc features from the time-frequency representation. The proposed method has been validated on a large set of signals from various difficult electrical loads. It proves successful in separating the series arcs from the perturbations produced by the load. Furthermore, it can easily be implemented for real-time processing.
Leprettre et al. (2002) studied this question.