Premium
Influence of model parameter estimation methods and regression algorithms on curing kinetics and rheological modelling
Author(s) -
Abliz Dilmurat,
Artys Tatjana,
Ziegmann Gerhard
Publication year - 2017
Publication title -
journal of applied polymer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.575
H-Index - 166
eISSN - 1097-4628
pISSN - 0021-8995
DOI - 10.1002/app.45137
Subject(s) - estimation theory , regression analysis , nonlinear regression , linear regression , model selection , curing (chemistry) , rheology , regression , algorithm , nonlinear system , linear model , model parameter , computer science , mathematics , biological system , statistics , materials science , physics , quantum mechanics , polymer chemistry , composite material , biology
The accuracy of the phenomenological curing and rheological models are strongly related to the estimated start parameters and selected regression algorithms. Considering the versatile methods for model start parameter estimation (model‐free vs. model‐fitting, dynamic vs. isothermal) and regression analysis algorithm (linear vs. nonlinear, single‐target vs. multi‐target), this paper investigates the theoretical basis and influence of these aspects on the model development process and model quality. The curing kinetics is modelled by model‐free and model‐fitting start parameters and different regression algorithms, followed by cross model validation at the final. The results showed that the different parameter estimation methods and evaluation algorithms have a remarkable influence on the final model parameters and its quality. The study shows the correlation between the different aspects and provides a basis for better selection of model parameter evaluation methods and regression algorithms for model development with improved quality and accuracy. © 2017 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2017 , 134 , 45137.