IMPROVED WIND SPEED PREDICTION USING VARIOUS NEURAL NETWORK MODEL

This work presents the feed forward neural models – Back Propagation Neural Network (BPN) model and Radial Basis Function (RBF) neural models employed to perform wind speed prediction and also the proposed approach of selecting number of hidden neurons embedded into the BPN and RBF models for performing wind speed prediction. Added to this contribution, it proposes hybrid neuro-fuzzy model for the considered application and as well the predicted wind speed employing this approach is compared with that of the basic neuro-fuzzy model i.e., Adaptive Neuro-Fuzzy Inference Systems (ANFIS). This work contribute to carry out effective wind speed prediction so that it enhances the wind system output for renewable resource power generation and the other side on the development of adaptable and scalable neural network architectures with fixed number of hidden layer neurons in BPN model, RBF model and ANFIS model. Considering all three proposed neural network models, the ANFIS network achieves the minimal mean square error resulting in improved wind speed prediction rate.