A Hybrid Long-Term Probabilistic Net Load Forecasting Approach Considering Renewable Energies Power in Smart Grids

With the growth and integration of distributed generation resources in smart grids, net load forecasting is of significant importance. A hybrid optimization method is proposed in this paper for probabilistic net load forecasting using neighborhood component analysis and solving regression problem with the aid of mini-batch LBFGS method. Net load forecasting is suggested in this paper trough forecast combination via adaptive network-based fuzzy inference system. The structure includes a combination of several long-term forecasts, including forecasts of load, the generation of a solar station, and the generation of a wind farm with wind turbines equipped with doubly-fed induction generator. Also, the net load forecasting and the relationship between errors of load, wind and solar predictions are studied in this paper. The simulation results of the proposed method and its comparison with Tao and quantile regression models show that mean absolute percentage error of load forecasting, and the forecasts of solar and wind generations improved by 0.947%, 0.3079% and 0042%, respectively which result to a decrease in net load forecasting error.