Prediction of wind speed with non-linear autoregressive (NAR) neural networks

In this study, the wind speed prediction model is created that gives a minimum error for different hidden layer neuron numbers and delay step numbers. Using the one-minute time series, the prediction of the next wind speed is performed with the NAR neural network model. The predicted values of wind speed obtained are compared with predicted values of wind speed obtained with filter methods. For different window functions and lengths, wind speed prediction is made using filters with different weight coefficients. For the number of hidden layer neurons is 14 and the number of delay steps is 10, MAE, MSE and RMSE values are calculated as 0.0315, 0.0019, 0.0445, respectively, with NAR neural network. It is seen that the proposed method for the wind speed dataset has a higher prediction performance than thefilter methods.