We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Identification of power quality events: selection of optimum base wavelet and machine learning algorithm.
- Authors
Hafiz, Faizal; Swain, Akshya; Naik, Chirag; Abecrombie, Scott; Eaton, Andrew
- Abstract
This study comprehensively investigates power quality (PQ) identification problem and proposes the optimum combination of base wavelet and machine learning algorithm (MLA) which would yield the highest classification accuracy. Although this problem has been studied by various researchers in the recent past, the selection of appropriate base wavelet and MLA, which would give better classification accuracy, have received comparatively less attention. This study bridges this gap by investigating the classification performance of 110 wavelets and 7 well-known MLAs across various noise levels using over 3500 PQ events generated as per IEEE Standard 1159. The results of this investigation demonstrate that the choice of base wavelet does significantly affect the classification performance. Further, it was observed that a single base wavelet does not provide optimum performance across all MLAs at various noise levels. In contrast, each MLA gives the maximum accuracy with a distinct base wavelet. The robustness of MLA against noise is studied which establishes that the simple MLAs, such as decision tree and Naive-Bayes, are more robust against noise compared to the other intricate MLAs. Finally, several recommendations are drawn for the selection of base wavelet and MLA which yields the best possible accuracy.
- Subjects
WAVELET transforms; MACHINE learning; CURVELET transforms; SUPPORT vector machines; SIGNAL processing
- Publication
IET Science, Measurement & Technology (Wiley-Blackwell), 2019, Vol 13, Issue 2, p260
- ISSN
1751-8822
- Publication type
Article
- DOI
10.1049/iet-smt.2018.5044