Enhancing wind power forecasting accuracy: A hybrid SNGF-RERNN-SCSO approach

Abstract

Forecasting wind power accurately is essential for optimizing energy management, improving grid stability. However, predicting wind speed and power generation is inherently challenging due to the intermittent and stochastic nature of wind patterns. The proposed hybrid system integrates the Surface Normal Gabor Filter (SNGF), recalling enhanced recurrent neural network (RERNN) and sand cat swarm optimization, named as SNGF-RERNN-SCSO approach. The SNGF efficiently reduces noise and refines wind data, while RERNN accurately predicts future wind speeds. The model’s computational efficiency is further enhanced by SCSO. The system gives an optimal solution with less calculation time by using the proposed technique. Then, the proposed approach is put into practice in the MATLAB platform and its execution is assessed with present strategies like Random Forest Algorithm (RFA), Recurrent Neural Network (RNN) and Giza Pyramid Construction. The proposed SNGF-RERNN-SCSO achieves lowest Mean Absolute Error (MAE) of 0.1 %, Mean Absolute Percentage Error (MAPE) of 2%, and Root Mean Square Error (RMSE) of 0.3. Furthermore, the proposed technique accomplishes the highest sensitivity of 98.06% maintaining the fastest execution time of 0.3 s. This emphasizes the higher accuracy and computational efficiency of the model, making it a robust and scalable solution for wind power forecasting.