Browse by author
Lookup NU author(s): Dr Pu Shi, Dr Wenxian YangORCiD
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
Wind energy has attracted more and more interest due to its clean, non-pollution and renewable features. In order to reduce wind turbine operation and maintenance costs, this research introduces an improved Back Propagation (BP) neural network technique for wind turbine drive train condition monitoring, especially for fault classification. In this approach, four types of fault signal, which are collected from main drive train components such as the main bearing, shaft, gearbox and generator, are used for fault classification. Firstly, in order to eliminate the effect of the difference of the orders of magnitudes, all the selected input signals are normalized. Secondly, to improve the BP neural network convergence efficiency and speed, improved learning algorithms are employed to train the proposed network. Thirdly, this paper introduces a new guideline to determine the hidden layer nodes. The test results show that the proposed network has real potential for wind turbine condition monitoring and fault classification.
Author(s): Shi P, Yang Wenxian, McKeever P, Ng C, Lee H
Editor(s): European Wind Energy Association
Publication type: Conference Proceedings (inc. Abstract)
Publication status: Published
Conference Name: EWEA Offshore 2015
Year of Conference: 2015
Print publication date: 12/03/2015
Online publication date: 12/03/2015
Acceptance date: 16/01/2015
Publisher: EWEA
URL: http://www.ewea.org/offshore2015/conference/programme/info2.php?id2=266&id=26%20&ordre=9#top