WIND POWER FORECASTING USING DEEP LEARNING TECHNIQUES

Abstract

The safety and stabilisation of the electricity network will be challenged as wind power’s share of the grid continues to rise, which will ultimately limit the scope of wind power development. For the wind energy industry to grow sustainably, forecasting is crucial. Power system dispatching may be done with confidence thanks to high-precision wind power forecasting. Since wind energy is intermittent and occurs over a range of time scales, increased wind power generation necessitates accurate wind forecasting. Numerous models have been developed for precise wind power prediction because of its stochastic character. Here, an accurate deep learning prediction model employing the LSTM (Long Short-Term Memory) technique is suggested.The model generates a forecast of wind power after being trained with real-time data from a wind farm. Two approaches, ARIMA and Random Forest, were compared in order to ascertain this model’s efficacy. The outcome demonstrates that the LSTM approach offers greater forecasting accuracy while requiring less MAPE. In Python software, the LSTM, Random Forest, and ARIMA models are implemented.