Melt index prediction by aggregated RBF neural networks trained with chaotic theory

Abstract

Melt index (MI) is considered as one of the most significant parameter to determine the quality and the grade of the practical polypropylene polymerization products. In this paper, the chaotic property of the melt index series is put forward for the first time, where its correlation dimension D2 is obtained to be 1.57, and the maximum Lyapunov exponent is 0.143. Through a fractional dimension and a positive maximum Lyapunov exponent, we demonstrate that the random nature of melt index can be explained as a chaotic phenomenon. Then, the phase space of the melt index series are reconstructed based on the obtained chaotic characteristics, and a novel RBF prediction model for MI prediction (RBF-chaos) are therefore further set up naturally to characterized its strong nonlinear and correlated relationships under chaotic theory. Furthermore, the detailed comparative researches between the RBF-chaos model and the other forecast models reported in the open literatures are carried out, and the research results show that the proposed chaos based neural network approach is superior to the previous models without considering chaotic characteristics.