This chapter presents the forecasting of wind power generation from wind turbines in Texas, using developed FONN models. The collected dataset contains 8760 samples, which include the essential parameters of wind speed, direction, air temperature, and pressure. Using both the single and multi-layer FONN models evaluates the efficiency of conventional and fractional-order activation functions in predicting wind power. From the results, it can be observed that the fractional-order activation functions have shown better performance, particularly fractional ArcTan and fractional LeCun Tanh, which achieve the highest accuracy in terms of \(R^2\) and MSE. Furthermore, this chapter addresses different scenarios with missing input data by employing Bi-LSTM for imputation, which further enhances forecasting accuracy. These findings highlight the potential of fractional-order neural networks in providing reliable power predictions for renewable energy systems.