The dWPF model
The dWPF model provides point-predictions for the wind power production of single wind farm or a whole area. It consists of a HML-RBFNN (short-term memory) and a complex structure of three RBFNN (long-tem memory). At the long-term memory, the Numerical Weather Predictions are grouped to three fuzzy sets. Depending on the defuzzification of each set, a different RBFNN is activated.
The HML-RBFNN and the above RBFNN structure are combined by the Recursive Least Squared method.
The dWPF model has evaluated in many cases with any type terrain with great performance.