Interactive Graphical User Interface (GUI) for Wind Speed Prediction Using Wavelet and Artificial Neural Network

Abstract

Large wind farms are being installed globally to meet power shortage and produce green power. The power output of the wind farm is dependent on the wind speed at any instant. In scheduling the generation and demand a few hours ahead, generators and distribution companies are required to project their schedules a priori to enable load dispatch centres to operate the grids reliably. A generation schedule is not easy with renewable sources especially wind due to the erratic nature of the wind speed. Hence it is vital to develop accurate models for wind speed prediction. In this paper hybrid models using neural networks are presented. Different pre-processing techniques for statistical input data of wind speed have been tried including wavelet decomposition. Novel training strategies for the neural network have been investigated. A generalised user friendly Graphical User Interface (GUI) tool has been developed for wind speed prediction wherein the user feeds historical data and the output of the GUI gives the predicted wind speed. The user has options to choose from different filtering techniques, training strategies and training algorithms for ANN. For the site location under consideration the mean percentage error obtained for the wind speed prediction was around 6%.