Open AccessLevenberg-Marquardt based MLP for Detection and Classification of Power Quality Disturbances
Author(s) -
Serge Raoul Dzondé Naoussi,
Jean Paul Ngon
Publication year - 2019
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.a2227.078219
Subject(s) - computer science , artificial neural network , fault (geology) , matlab , multilayer perceptron , power (physics) , perceptron , levenberg–marquardt algorithm , quality (philosophy) , artificial intelligence , generator (circuit theory) , pattern recognition (psychology) , physics , quantum mechanics , seismology , geology , operating system , philosophy , epistemology
In recent years, power quality (PQ) has become animportant issue for utilities and users. In order to improve PQ, amethod for detecting and classifying power quality disturbances(PQDs) is proposed. Hence in addition to identifying thedisturbance signals, the proposed method is able to determine itstype when occurring. This approach is based on Multilayerperceptron and Levenberg-Marquardt training rule. It is inspiredby the desire to take advantage of the parallelism inherent toneural networks in view of hardware implementation usingreconfigurable chips. The inputs of these networks are thesamples obtained on the power grid in various conditions. Theproposed method is tested for sags and swells. To classify thedisturbances, the neural architectures have been generalized andconfigured according to the number and type of disturbances tobe treated. To validate and test the proposal, a grid model wasbuilt with a three-phase fault generator under Matlab / SimulinkR2017a. After comparing the results with those obtained bycertain methods in the literature, the proposal proves to be anefficient and reliable tool for monitoring PQ. In fact it has thesmallest mean square error and a highperformance withprecision of 96%.