Forecasting of Wind Speed and Power for Poland Using Prominent Variants of FFNN Tuned Through MHPSOBAAC-x

production based on conventional fossil fuels, which exert a
considerable environmental burden. This dependency can be
mitigated through the increased utilization of Renewable Energy
Resources (RERs), which offer both environmental and economic
advantages. However, one of the main challenges associated with
RERs, particularly wind energy, is their frequent installation at
locations remote from load centers, which poses difficulties in
efficient energy transmission. In this study, wind speed and power
have been forecasted using three types of artificial neural networks:
Feedforward Neural Network (FFNN), Cascaded Forward Neural
Network (CFNN), and improved Feedforward Neural Network (idFFN). The dataset used in this work was obtained from the Global
Wind Atlas and covers the Silesian Region of Poland for the year
2023. Data from January to June were used to train the neural
networks using a modified version of the Metaheuristic Hybrid
Particle Swarm Optimization with Bat Algorithm Acceleration
Coefficients (MHPSO-BAAC-x). The trained models were then
employed to predict wind speed and power for the period from July
to December 2023. To assess the accuracy of the forecasts, the
following statistical error metrics were applied: Mean Absolute Error
(MAE), Mean Absolute Percentage Error (MAPE), and Root Mean
Square Error (RMSE). The predicted values were subsequently
compared with the actual data for the same period to evaluate the
effectiveness of the models. The id-FFN model demonstrated the best
performance, achieving values of 0.0262 m/s for MAE, 2.62% for
MAPE, and 0.0152 m/s for RMSE, confirming its high precision and
reliability. The obtained results suggest that the developed
forecasting system has the potential to support future planning and
integration of wind power stations into the national power system.