No abstract

With the rapid growth of wind power penetration into modern power grids, wind speed forecasting plays an increasingly significant role in the planning and operation of electric power and energy systems. However, the existing wind speed forecasting methods are modeled as black boxes, which are very complicated and cannot be written down explicitly due to the complex fluctuation characteristics of wind speed series. To this end, this study proposes a novel direct method based on an explainable neural network (xNN) for deterministic and probabilistic wind speed forecasting. It can theoretically extract the nonlinear mapping features in wind speed, thereby providing a clear explanation of the relationship between the input and the output of the forecasting model. Then, the uncertainties in wind speed are statistically synthesized via the kernel density estimation method. Finally, we use wind speed data from real wind farms in Belgium to verify the feasibility and effectiveness of the proposed method. The simulation results demonstrate that it is not only able to accurately extract the non-stationary feature in the wind speed series but also superior to other benchmark algorithms in prediction accuracy. Therefore, the proposed method has a high potential for practical applications in real electric power and energy systems.