|
Descripción
|
|
|---|---|
| The changes in the local strain field or the global stiffness caused by the occurrence of a damage in a wind turbine blade are usually very small and can be faded by external effects (i.e. load conditions, environmental conditions, etcetera). Only by comparing strain measurements at several points across the blade some information about damage may be unveiled. This task can be achieved in an automatic way by means of pattern recognition techniques based on the strain field. This is the basis of the strain measurements data-driven models. Nowadays, one of the most promising techniques for obtaining strain measurements is distributed sensing. By means of an Optical Backscatter Reflectometer (OBR) it is possible to get the strain along a plain optical fiber with a good spatial resolution and strain accuracy (depending on a combination of related measurement parameters). The strain can be gathered ?continuously? along the fiber unlike as with the use of strain gauges or fiber Bragg gratings where only discrete strain can be obtained. Since distributed sensing allows to use high amount of virtual sensors it is possible to study the damage detection sensitivity as a function of the number of sensors. In that way, it would be possible to estimate an optimum sensor network architecture. Within the context of this paper two different pattern recognition approaches have been studied as mechanisms to quantify the damage detection sensitivity. Two main objectives were exploited: first, reduce the dimensionality of the problem under study, and second extracting features hidden within the data. The first approximation consisted in a linear PCA model and the second one consisted in hierarchical nonlinear PCA model (h-NLPCA). For both, an outlier analysis was performed by means of the Q index and the T 2 index and their associated damage thresholds. An experimental validation of the technique is discussed in this paper. A 13.5 meters long wind turbine blade was used and several strain measurements for a healthy state and two damaged conditions were compared | |
|
Internacional
|
Si |
|
Nombre congreso
|
6th World Conference on Structural Control and Monitoring |
|
Tipo de participación
|
960 |
|
Lugar del congreso
|
Barcelona, España |
|
Revisores
|
Si |
|
ISBN o ISSN
|
|
|
DOI
|
|
|
Fecha inicio congreso
|
15/07/2014 |
|
Fecha fin congreso
|
17/07/2014 |
|
Desde la página
|
|
|
Hasta la página
|
|
|
Título de las actas
|
Proceedings of the 6th World Conference on Structural Control and Monitoring |