Security Constrained Unit Commitment Based Load and Price Forecasting Using Evolutionary Optimized LSVR

In this paper, a local predictor approach based on proven powerful regression algorithm which is support vector regression (SVR) combined with space reconstruction of time series is introduced. In addition, real value genetic algorithm (GA) has been utilized in the proposed method for optimization of the parameters of the SVR. In the proposed approach, the embedding dimension and the time delay constant for the load and price data are computed firstly, and then the continuous load and price data are used for the phase space reconstruction. Subsequently, the reconstructed data matrix is subject to the local prediction algorithm. Then the forecasted loads and price are fed into IEEE 30 bus test system for security constraint unit commitment to show the reactions of unit commitment to load and price forecasting errors. The proposed model is evaluated using real world dataset. The results show that the proposed method provides a much better performance in comparison with other models employing the same data.