Performance Evaluation of Different ANN Models for Medium Term Wind Speed Forecasting

Abstract

Wind power generation is characterized by its variability and uncertainty in the wind speed. Thus, the integration of wind farms to utility grids has several impacts on the optimum power flow, transmission congestion, load dispatch, economic analysis, and electricity market clearing prices. Due to the irregular nature of wind power production, accurate prediction of wind speed poses a major challenge to researchers. Wind speed of a wind farm is affected by conditions of the environment in which the wind farm is built, such as temperature, humidity, dew point, atmospheric pressure and wind direction. In this paper, five ANN techniques namely FFBP, CFBP, PNN, GRNN and KNN are considered to predict the wind speed using MATLAB. The feasibility of the proposed techniques is evaluated using the performance measures such as MSE, MAPE and linear regression and it is observed that GRNN is superior amongst the other methods that are used.