A Neural Network Retraining Approach for Process Output Prediction

摘要

This paper is a report that describes a feedforward neural network architecture and an application of retraining algorithm in order to forecast relevant process variables representative for glass manufacturing, provided by EUNITE Competition 2003. The main purpose is to establish an optimum feedforward neural architecture and a well suited delay vector for data forecasting. The artificial neural networks (ANNs) ability to extract significant information provides valuable framework for the representation of relationships present in the structure of the data. The evaluation of the output error after the retraining of an ANN shows us that this procedure can substantially improve the achieved results.