Comparative investigation of short-term wind speed forecasting models for airborne wind turbines

Airborne wind turbine technology is rapidly growing in purpose to overcome limitation of wind turbines working at low altitude. The high-altitude wind is strong to efficient power generation. Under varying wind conditions, wind forecasting in real time is necessary to be implemented for flight stabilization and power generation. This study is to investigate three widely-used forecasting models for short-term wind speed prediction from wind measurement data, that is persistence model, autoregressive moving average (ARMA) model, and artificial neural network (ANN). It is found that accuracy of persistence model dramatically decreases as time horizon increases; nevertheless, the persistence model is the simplest algorithm for implementation. The ARMA model and ANN yields significant accuracy of wind speed forecasting compared with the persistence model. The overall mean absolute errors of ARMA and ANN are 19.78% and 22.69% lower than the persistence method, respectively. The lowest errors are found in ANN for most cases of time horizon lengths. Unlike ANN, setup of ARMA model is systematically in hand. A few time horizons can be recommended for short-term wind speed forecasting for airborne wind turbine.