NARX Based Short Term Wind Power Forecasting Model

Nowadays, with the growing needs of the consumers there is a huge demand for the electric power but the fuel reserves are also depleting at the same pace. So, this has created the need to depend up on the renewable energy resources to meet the required power demand.  Since the power generated through renewable resources is eco friendly in nature and distributed, this is an added advantage. Of all the renewable energy resources solar and wind plays the most crucial part in the power generation because of their wide spread availability. But the wind energy is volatile and intermittent by nature, due to this interconnecting the power generated to grid becomes a hectic task. So in this paper a wind power forecasting model with the help of artificial neural networks (ANN) is developed so that the wind power can be forecasted well in progress, which helps in maintaining and operating grid interconnection and also scheduling of units. The developed model is based on the non-linear auto regressive with exogenous input (narx) tool which trains the ANN for the time series. The input parameters taken into consideration are wind speed, temperature, pressure, air density and the output parameter is generated power. The required data is collected from the Energy Department of KLUniversity, Andhra Pradesh which consists of 720 hours data from that 672 hours data is used for training and 48 hours data is used for prediction. Mean square error and root mean square error are calculated from the predicted and known results. DOI:  http://dx.doi.org/10.11591/telkomnika.v15i1.8070