Evolving feedforward neural network for harmonics signature identification

Authors:
Win Siau Ng;Dipti Srinivasan;Ah Choy Liew
Affiliations:
Electrical and Computer Engineering Department, National University of Singapore;Electrical and Computer Engineering Department, National University of Singapore;Electrical and Computer Engineering Department, National University of Singapore
Venue:
Design and application of hybrid intelligent systems
Year:
2003

Citing 2
Cited 0

What is evolutionary computation?

IEEE Spectrum
A new evolutionary system for evolving artificial neural networks

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.01

Visualization

Abstract

This paper presents artificial neural network (ANN) evolved by genetic algorithm (GA) for signature identification of electrical devices based on the power harmonics generated. Initially, the GA was used to evolve the weights and biases of the single-hidden-layer perceptron ANN. To improve the ANN's performance, the GA's role was modified to evolve only the initial weights of the ANN while backpropagation training was employed for convergence towards the minima. Aetnal data obtained from several devices was used for evaluation and testing of this neural network for signature identification. Very promising results, with classification accuracy above 98% have been obtained with this hybrid approaches.