Chapitre D'ouvrage Année : 2017

A new wavelet−ANN approach based on feature extraction for a FAST wind turbine model diagnosis system

Résumé

This work presents a method to increased the fault detection accuracy in wind turbine system using a combination of Wavelet, Principal Component Analysis (PCA), Parseval’s theorem and Neural Networks. Preprocessing, feature extraction and classification rules are three crucial issues for fault detection. To employ this issues, a two stages hybrid approach is used. The first stage is composed of preprocessing and feature extraction steps, where wavelet transform is exploited for preprocessing residual signals while two tools are used and compared to extract features based on PCA and Parseval’s theorem. During the classification stage, the Artificial Neural Network (ANN) is explored to achieve a robust decision in presence and absence of faults. The approaches are applied on a FAST wind turbine system.

Fichier non déposé

Dates et versions

hal-01918830 , version 1 (12-11-2018)

Identifiants

  • HAL Id : hal-01918830 , version 1

Citer

Tahani Bakir, Boussaid Boumedyen, Peter Odgaard, Mohamed Naceur Abdelkrim, Christophe Aubrun. A new wavelet−ANN approach based on feature extraction for a FAST wind turbine model diagnosis system. Faouzi Derbel and Nabil Derbel and Olfa Kanoun. Power Systems & Smart Energies, 7, De Gruyter, pp.155-178, 2017, Advances in Systems, Signal and Devices, 978-3-11-047052-9. ⟨hal-01918830⟩
103 Consultations
0 Téléchargements

Partager

  • More