Prediction of glass transition temperatures of aromatic heterocyclic polyimides using an ANN model

Abstract

Aromatic heterocyclic polyimides are used extensively in industries for their excellent mechanical properties, high glass transition temperatures (Tg), and so on. A quantitative structure–property relationship (QSPR) model was developed to predict the Tg values with 54 aromatic heterocyclic polyimides by using an artificial neural network (ANN) back‐propagation algorithm. Fifty‐four aromatic heterocyclic polyimides were randomly divided into a training set (36) and a test set (18). Three molecular descriptors (the connectivity index X1A, the topological descriptor PW3, and the 3D‐MoRSE descriptor Mor09e) were selected to produce the mode. Simulated with the final optimum ANN model with 3‐3‐1 structure, the results show that the predicted Tg values are in good agreement with the experimental ones, with the root mean square errors (RMSEs) of 12.4 K (R = 0.935) and 16.4 K (R = 0.937) for the training set and the test set, respectively. POLYM. ENG. SCI., 50:1547–1557, 2010. © 2010 Society of Plastics Engineers

Prediction of elongation at break for linear polymers

“…For these reasons, it becomes a challenge to generate a reliable associated dataset too. One of the earliest and most widely studied of polymer properties has been the T g , and good prediction results were obtained from synthetic models [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24]; in contrast, the mechanical properties of polymers have scarcely been explored. Seitz [25] developed semi-empirical and empirical relationships so as to estimate the mechanical properties of polymeric materials from the molecular weight, van der Waals volume, the length and number of rotational bonds in the repeat unit, besides the T g of the polymer.…”

Prediction of polymer properties using infinite chain descriptors (ICD) and machine learning: Toward optimized dielectric polymeric materials

“…In the past 30 years, numerous attempts have been made to predict polymer properties, where the glass transition temperature may be the most studied one . Studies on the dielectric constant and refractive index have also been reported .…”

“…Even though studies such as the one described in Katrizky's 1996 paper considered the extension of descriptor values for long chains using a numerical treatment, most of the models simply consider a repeat unit with different end‐cap atoms as a complete molecular representation. Monomers with carbons as end‐cap atoms, dimers, and ring‐like dimers have been used as ways of incorporating features of the polymer chemical environment. With these models, two theoretical issues arise: If a descriptor value is an “extensive” property that varies with the scale of the local molecular representation (e.g., the monomer and dimer give different descriptor values), the information contained in this descriptor is contaminated by the choice of the molecular representation.…”

“…may be the most studied one. 1,[4][5][6][7][8][9][10][11][12][13][14][15][16][17] Studies on the dielectric constant and refractive index have also been reported. [16][17][18][19][20][21] Le et al 22 did a comprehensive review on the QSPR modeling for material properties in 2012.…”

Multiobjective optimization design of heating system in electric heating rapid thermal cycling mold for yielding high gloss parts

“…Artificial neural network (ANN) can be considered as an alternative to the polynomial regression. Among various kinds of ANN models, the error back propagation (BP) feed‐forward neural network is the most widely utilized one, and it has been used for modeling in many processes . Moreover, stochastic search procedure based on the GA can be an efficacious manner for process optimization.…”