Prediction error feedback for time series prediction: a way to improve the accuracy of predictions
Ryad Zemouri, Paul Ciprian Patic · 2010
Many applications using recurrent Artificial Neural Network (ANN) for system output prediction do not feedback the previous prediction error to perform the calculation of a new prediction output. The feedback of this prediction error is re-used in two different ways in the neural network. The prediction tests have been carried out on the Mackey-Glass chaotic time series, Logistic Map time series and Box-Jenkins furnace data.