Using neural networks to detect wind turbine functioning conditions

GINO IANNACE, Giuseppe Ciaburro, Amelia Trematerra


Wind has always represented a source of energy for human being. Currently, companies all over the world invest huge capital to build wind farms with the aim of obtaining the maximum possible economic return. Therefore, the identification of sites with the greatest windiness is necessary. These sites often reside in rural areas where the environmental impact of wind turbines, especially the noise impact, is significant. In this study, measurements of the noise emitted by several wind turbines located in South Italy were made. A selected range of the average spectral levels in a 1/3 octave band was used to identify the wind turbine operating conditions. A model based on neural network for detection operating conditions of the wind turbines was hence developed and applied. The results show the high accuracy of the forecast and identification model and suggest the adoption of this tool for several other applications.


Artificial neural network, low-frequency sound, wind turbine noise, feature selection, random forest

Full Text: PDF


  • There are currently no refbacks.

Copyright © 2011-2018 AUSMT ISSN: 2223-9766