A prediction method for gas emission based on RBF with grey correlation analysis

Abstract

A rolling method of gas emission based on RBF neural networks is improved. In this method, a part of fixed-length data is selected for the prediction, new data are added continuously to the input sequence, and old data are removed, thereby developing the rolling prediction model. The diversified factors of gas emission analyzed have grey correlation. As a result, the model designed using this method can generalize well. The simulation results also show that the improved rolling prediction model applied in gas emission prediction has reliable accuracy and a good convergence rate.