Waveform segmentation for intelligent monitoring of power events

► Kalman- and tensor-based algorithms for waveform segmentation are compared. ► We analyze their performance using real events generated by four different causes. ► Faults inserted close to zero-crossing introduce large error in Kalman algorithms. ► Algorithms can adequately segment Animal/cable events, tree ones only by tensor-based. ► Algorithms have more difficulty segmenting fuse and transformer events.