Multivariate wind power forecast using artificial neural network

Power generations from renewable sources of energy like solar and wind is catching up rapidly. There is a dire need for forecasting the generation in order to have better load scheduling as the generation is uncertain because the weather is erratic and the generation depends on a lot of factors. Therefore with greater penetration of renewable sources in power generation, the focus is shifting towards generation forecasting. This paper proposes predictive models for wind power generation using non-linear auto regressive neural network. Three multivariate models are developed for a day ahead prediction of wind power generation. A comparative study is done on the errors and it is found that wind speed is highly dependent on wind direction. A model with wind speed and wind direction as inputs gives better prediction.