Regularization of short term load forecasting neural models

The knowledge of loads’ future behavior is very important for decision making in power system operation. During the last years, many load models have been proposed, and the neural ones have presented the best results. One of the disadvantages of the neural models for load forecasting is the possibility of excessive adjustment of the training data, named overfitting, which degrades the generalization capacity of the estimated models. This problem can be tackled by using regularization techniques. This paper shows the application of some of these techniques to short term load forecasting. Index Terms — Short-term load forecasting, artificial neural networks, regularization techniques, Bayesian training, gain scaling, support vector machines.