Forecasting of wind speed using ANN, ARIMA and Hybrid models

Wind energy is one of the sustainable and renewable energies which have been the center of attention in industrial communities. The wind energy that is obtainable is arbitrary. So available wind power can't be known ahead in time. For the system organization as well as energy dispatching, a wind farm operator should know the power from wind in advance. Wind speed forecasting helps to minimize the unreliability in wind power forecasting. It enables for more desirable grid organization and incorporation of wind with power systems. It also helps to reduce the imbalance charges and penalties. For scheduling the power, the wind farm operator must at least know the wind speed data of one day ahead and for confirmation, he wants to know the wind speed data of one hour ahead. This paper is about the wind speed forecasting for different time horizons in three different sites in Tamil Nadu, India (Dharapuram, Kayathar, Nollur) using three different models such as Artificial Neural Network (ANN) method, Auto Regressive Integrated Moving Average Method (ARIMA) and Hybrid method, which is the combination of ARIMA and ANN and the comparison of their results. The wind speed data for three years with 1 hour time block is taken as inputs for the considered model. For predicting the accuracy of the model, numerical error evaluation methods such as Mean Absolute Percentage Error (MAPE), Mean Square Error (MSE), and Mean Absolute Error (MAE) were used. The outcome showed that the forecasted error for hybrid model is lesser than artificial neural network and autoregressive moving average model separately.