Improved Multilayer Perceptron Neural Network Input Parameters for Load Forecasting

C. Senabre, S. Valero, M. Ortiz, and A.Gabaldón (Spain)


Short-term load forecasting, MLP neural networks, electricity market, demand management.


Quick and accurate load forecasting is very important for power system operation. Load forecasting is of considerable interest to electric power utilities. The objective of the research has been to analyze different parameters from training of a Multilayer Perceptron Neural Network (MLP), in order to obtain the best results of short-term load forecasting. Historical global load data of Spain and temperature data have been considered in this study. The authors define a pattern of input data and training parameters for the best results. This technique allowed to obtain the global demand curve for 24 hour periods (hour to hour). In order to validate the model, several error indices were assigned through the comparison of the results with the real known curves. Summing up, the research establishes a tool that helps in the decision making, forecasting the short-term global electric load demand curve.

Important Links:

Go Back