SHORT-TERM WIND SPEED ESTIMATION WITH ANN BASED HYBRID OPTIMIZATION METHODS IN ÇUKUROVA REGION

Abstract

Short term wind speed estimation has a great importance in the renewable energy field. Because the continuous change of the wind energy depending on the wind speed causes the production/consumption imbalance in the networks. In order to prevent instability of the voltage and frequency changes, the balance between electricity production and consumption in the network must be provided continuously. Moreover, the installation costs of wind energy systems are relatively high so wind speed estimation studies are significant to minimize wind energy losses in the future. Nonlinear changes in the wind speed affect the modeling time and accuracy of the wind energy systems. Therefore, a short-term wind speed estimation method based on artificial neural network is proposed. In this study, a short-term wind speed estimation is performed by using the Artificial Neural Network (ANN) and ANN-based hybrid methods such as ANN-Harmony Search (ANN-HS) and ANN-Invasive Weed Optimization (ANN-IWO). Input parameters are obtained by using meteorological measurement stations. These parameters are the ambient temperature, wind speed and humidity in the Çukurova Region, Adana/Turkey. ANN-based hybrid estimation methods are preferred because of its success in modeling nonlinear sequences such as wind speeds. These powerful methods used in estimation problems are combined and presented as a new hybrid approach in order to reduce the errors. The errors in the estimations are determined by Root Mean Square Error values (RMSE) so the performance estimation results between the hybrid methods and ANN methods are presented in detail.