Short-term power load forecasting by neural network with stochastic back-propagation learning algorithm

Abstract

In this paper, a short-term power load forecaster based on a neural network with stochastic back-propagation learning algorithm is developed. This modified learning rule can effectively help the load forecaster escape from a local minimum while it is trained. Consequently, the proposed load forecaster has more accurate prediction in forecasting operation. As a comparison, the same experiments are also performed by using a neural network with a traditional back-propagation learning rule which has constant learning rate and momentum.