Wind Power Forecasting Using Hybrid ARIMA-ANN Technique

Abstract

The wind power forecasting along with the prior knowledge of wind speed has become very important for the efficient functioning of wind power generation and effective management of risk and revenue. Several single approach models are there for forecasting of wind power, i.e., ARIMA, support vector machine (SVM), artificial neural networks (ANN), extreme learning machine (ELM), etc., but hybridization of these models is considered as an effective alternative for forecasting. In the proposed work, the hybridized model combining ARIMA and artificial neural network (ANN) is presented in order to provide a better prediction of wind power. The wind speed data of Denmark is used for evaluation of the proposed model. From the result obtained, it becomes evident that the hybridization of ARIMA and ANN is better in forecasting the wind power as compared to the two models working separately for wind power forecasting.