Improved characterization of Multi-Stage Voltage Dips based on the Space Phasor Model
Artikel i vetenskaplig tidskrift, 2018

This paper proposes a method for characterizing voltage dips based on the space phasor model of the three phase-to-neutral voltages, instead of the individual voltages. This has several advantages. Using a K-means clustering algorithm, a multi-stage dip is separated into its individual event segments directly instead of first detecting the transition segments. The logistic regression algorithm fits the best single-segment characteristics to every individual segment, instead of extreme values being used for this, as in earlier methods. The method is validated by applying it to synthetic and measured dips. It can be generalized for application to both single- and multi-stage dips.

Electric power distribution

Machine learning algorithms

Voltage dips

Clustering algorithms

Electric power transmission

Logistic regression

Power quality


Azam Bagheri

Lulea tekniska Universitet

Mathias Bollen

Lulea tekniska Universitet

Irene Yu-Hua Gu

Elektroteknik, Signalbehandling och medicinsk teknik, Signalbehandling

Electric Power Systems Research

0378-7796 (ISSN)

Vol. 154 319-328




Elektroteknik och elektronik