Browse by author
Lookup NU author(s): Dr Jie ZhangORCiD, Emeritus Professor Julian Morris
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
A type of wavelet neural network, in which the scale function is dopted only, is proposed for non-linear dynamic process modelling. Its network size is decreased significantly and the weight coefficients can be estimated by a linear algorithm. The wavelet neural network holds some advantages superior to other types of neural networks. First, its network structure is easy to specify based on its theoretical analysis and intuition. Secondly, network training does not rely on stochastic gradient type techniques and avoids the problem of poor convergence or undesirable local minima. The excellent statistic properties of the weight parameter estimations can be proven here. Both theoretical analysis and simulation study show that the identification method is robust and reliable. Furthermore, a hybrid network structure incorporating first-principle knowledge and wavelet network is developed to solve a commonly existing problem in chemical production processes. Applications of the hybrid network to a practical production process demonstrates that model generalization capability is significantly improved.
Author(s): Huang D, Jin Y, Zhang J, Morris AJ
Publication type: Article
Publication status: Published
Journal: Chinese Journal of Chemical Engineering
Year: 2002
Volume: 10
Issue: 4
Pages: 435-443
Print publication date: 01/08/2002
ISSN (print): 1004-9541
ISSN (electronic): 0892-0370
Publisher: Chemical Industry Press
