Chapter

Assessment and Simulation Tools for Sustainable Energy Systems

Volume 129 of the series Green Energy and Technology pp 247-266

Date:

Artificial Neural Network Based Methodologies for the Estimation of Wind Speed

  • Despina DeligiorgiAffiliated withDepartment of Physics Division of Environmental Physics and Meteorology, University of Athens Email author 
  • , Kostas PhilippopoulosAffiliated withDivision of Environmental Physics-Meteorology, Department of Physics, University of Athens
  • , Georgios KouroupetroglouAffiliated withDepartment of Informatics and Telecommunications, Division of Signal Processing and Communication, University of Athens

* Final gross prices may vary according to local VAT.

Get Access

Abstract

Recent advances in artificial neural networks (ANN) propose an alternative promising methodological approach to the problem of time series assessment as well as point spatial interpolation of irregularly and gridded data. In the field of wind power sustainable energy systems ANNs can be used as function approximators to estimate both the time and spatial wind speed distributions based on observational data. The first part of this work reviews the theoretical background, the mathematical formulation, the relative advantages, and limitations of ANN methodologies applicable to the field of wind speed time series and spatial modeling. The second part focuses on implementation issues and on evaluating the accuracy of the aforementioned methodologies using a set of metrics in the case of a specific region with complex terrain. A number of alternative feedforward ANN topologies have been applied in order to assess the spatial and time series wind speed prediction capabilities in different time scales. For the temporal forecasting of wind speed ANNs were trained using the Levenberg–Marquardt backpropagation algorithm with the optimum architecture being the one that minimizes the Mean Absolute Error on the validation set. For the spatial estimation of wind speed the nonlinear Radial basis function Artificial Neural Networks are compared versus the linear Multiple Linear Regression scheme.