Application of Rough and Fuzzy Set Theory for Prediction of Stochastic Wind Speed Data Using Long Short-Term Memory

Despite the great significance of precisely forecasting wind speed for development new and clean energy technology stable grid operators, stochasticity makes prediction a complex challenging task. For improving security economic performance power grids, accurate short-term is crucial. In this paper, deep learning model (Long Short-term Memory (LSTM)) has been proposed prediction. Knowing that time series nonlinear stochastic, mutual information (MI) approach was used to find best subset from data by maximizing joint MI between target output. To enhance accuracy reduce input characteristics uncertainties, rough set interval type-2 fuzzy theory are combined in model. Wind an international airport station southern coast Iran Bandar-Abbas City as original dataset optimized Based on statistical results, LSTM (RST-LSTM) showed better than LSTM, well traditional neural networks, with lowest error training testing datasets different horizons. The suggested can support optimization control smooth procedure system. results confirm superior capabilities techniques forecasting, which could also inspire applications meteorology assessment.