Real-time diagnosis and fault detection for the monitoring of the wind turbine

The present paper addresses fault signature analysis for fault detection and isolation problems. The aim of this article is to create a procedure for detecting and locating abnormal behaviour in an operating wind turbine before possibly catastrophic failures. Each of the three fault classes (bias, incipient fault, and intermittent fault) is successfully studied. Wind energy systems with a doubly fed induction generator example are illustrated to show the effectiveness of the proposed diagnosis method. An experimental benchmark emulating the behaviour of a wind turbine is used to confirm the approach. Finally, real time tests are performed to allow real time validation. (C) 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.4757624]