Nonlinear Semi-Parametric Modeling Mothed Based on GA-ANN

DUAN Bin,LIANG Jun,FEI Zheng-shun,YANG Min,HU Bin
DOI: https://doi.org/10.3785/j.issn.1008-973x.2011.06.003
2011-01-01
Abstract:Nonlinear se-mi-parametric models are introduced for industrial process modeling to improve the modeling accuracy by taking the advantages of both parameter and non-parameter models.The modeling methodology and structure of nonlinear semi-parametric modeling are proposed based on the genetic algorithm and the neural network,and the cross-loop iterative algorithm procedures are also introduced for estimating the parameters of both the parametric and non-parametric parts.Then,the design of neural network and the genetic algorithm are investigated,which increase the elite preserving strategy,enhance the memory function,propose an innovative fitness calculation method,and improve the crossover and mutation strategy.The on-site industrial data of polyethylene plant is used to demonstrate the effective of this method.The result shows that the proposed approach is more accurate in prediction than the conventional parametric models and can better track the variation of the process.
What problem does this paper attempt to address?