Dynamic modelling of a paper making process based on bilinear system modelling and genetic neural networks

Abstract

The dynamic modelling of the wet end of the paper machines has been recognised as a challenging problem due to its nonlinear, complex, time-varying, time-delayed, and multivariable interactive properties. This paper presents a methodology based on bilinear system modelling and multilayer perceptron (MLP) neural network for modelling of such a complex system. Genetic algorithm (GA) search and optimisation technique is proposed to train the neural network weights. This logical combination has advantages of both physical and genetic neural modellings.