Keywords
power quality disturbance;Markov translate field;visualization;dense convolutional networks;channel attention mechanism;classification and recognition
Abstract
Aiming at the problems of complex process and insufficient refinement of artificial feature selection in traditional power quality disturbances (PQDs) classifier, a new PQD recognition method based on Markov transition field visualization and improved DenseNet is proposed. Firstly, the one-dimensional PQD signal is mapped into a two-dimensional image by MTF. Then, the image is input into an improved DenseNet with a new channel attention mechanism. Finally, the network is trained to extract features from a large number of samples by itself, so as to realize the correct recognition of PQD signals. The example results show that: in the case of no noise and signal-to-noise ratio of 20dB and 30dB, the proposed improved DenseNet can effectively overcome the shortcomings of traditional methods, such as strong subjectivity of feature selection and poor anti-noise performance. It can better extract the feature information of complex PQD, and has a high recognition rate for complex PQD.
DOI
10.19781/j.issn.1673-9140.2024.04.012
First Page
102
Last Page
111
Recommended Citation
SHI, Shuai; CHEN, Ziwen; HUANG, Dongmei; HE, Qi; SUN, Yuan; and HU, Wei
(2024)
"An identification method based on MTF visualization and improved DenseNet for power quality disturbances,"
Journal of Electric Power Science and Technology: Vol. 39:
Iss.
4, Article 12.
DOI: 10.19781/j.issn.1673-9140.2024.04.012
Available at:
https://jepst.researchcommons.org/journal/vol39/iss4/12