Forecasting Wind Speed Using the Proposed Wavelet Neural Network

Wind energy is one of the speedy processing technologies in generation industry and most economical methods electrical power generation. For reliability system, it wanted to improve highly appropriate wind speed forecasting methods. The wavelet transform a powerful mathematical technique that converts an analyzed signal into time-frequency representation. This has proven useful nonstationary time series forecasting. aims this study are propose function by derivation quotient from two different Lucas polynomials, as well comparison between artificial neural network (ANN) wavelet-artificial (WNN). We used proposed wavelet, Mexican hat, Morlet, Gaussian, Haar, Daubechies, Coiflet data using continuous (CWT). MATLAB software was implement CWT ANN. models were applied meteorological field forecast daily collected directorate Sulaymaniyah which city located Kurdistan region Iraq for period (Jan. 2011–Dec. 2020). Five performance criteria during calibration validation, root mean square error ( R mathvariant="normal">M mathvariant="normal">S mathvariant="normal">E ), id="M2"> absolute percentage id="M3"> mathvariant="normal">A mathvariant="normal">P , id="M4"> coefficient determination id="M5"> R 2 evaluated. When studying, analyzing, comparing these models, results concluded wavelet-ANN best result id="M6"> = 0.00072 id="M7"> 0.02683 id="M8"> 2.32400 id="M9"> 0.99983 .