An analytic evaluation of frequently used artificial neural network algorithms based on the electricity load forecast of Antalya province

In this study, Artificial Neural Networks (ANN), using Turkey's Antalya Province is located in the southern region of electric load forecasting, based algorithms have been evaluated. In this study, the electrical model of MATLAB neural network toolbox and a feedback network is designed with three floors and has been resolved. Electrical (E), temperature (S), humidity (N), pressure (B) and population (P) values to the input data used in the model between the years of 2001-2011. In the model, preparing the grant application ESBN, ESN, EBN, EBS and ESBP input data was used. Thirty Algorithms predicted values were obtained with the help Levenberg-Marquardt (LM), Gradient-Descent (GD), Resilient Propagation (RP), Gradient-Descent-Momentum (GDM), Gradient-Descent-Adaptive (GDA) and Gradient-Descent-Adaptive-Momentum-Rate Backpropagation (GDX). Root mean square error (RMSE) was estimated with the values obtained. RMSE, the number of iterations, such as processing time criteria on which to base an Analytical Hierarchy Process (AHP) model is developed. ANN commonly used algorithms are evaluated. In this study, ANN model was run with three or more inputs and ANN algorithms commonly used case study of Antalya province many-criteria decision-making tool was evaluated by the AHP method.