Prediction of thermal conductivity of polymer-based composites by using support vector regression

Wang, GuiLian; Cai, Can; Pei, J.F.; Zhu, XingJian

doi:10.1007/s11433-011-4319-8

Cited by 11 publications

(8 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SVR has been proved to exhibit a number of significant advantages such as excellent learning performance of small samples, good generalization ability, small errors, high calculation accuracy, etc 15. At present, it has become a focus in machine learning research and is extensively employed in a wide range of real‐world problems 16–30…”

Section: Methodsmentioning

confidence: 99%

Modeling and Predicting the Glass Transition Temperature of Polymethacrylates Based on Quantum Chemical Descriptors by Using Hybrid PSO‐SVR

Pei

Cai

Zhu

et al. 2012

Macro Theory & Simulations

View full text Add to dashboard Cite

Based on six quantum chemical descriptors (jL-1.356j, E total , qC6, a, q À , and E therm ), the hybrid PSO-SVR is proposed to establish a model for predicting the glass transition temperature (T g ) of 37 polymethacrylates. The prediction performance of SVR was compared with those of reported MLR and ANN models. The results show that the RMSE, MAPE, and R 2 calculated by SVR are superior to those achieved by MLR or ANN model for the identical training set and test set. This investigation reveals that the SVR model is more suitable to be used for prediction of the T g values for unknown polymethacrylates possessing similar structure than the conventional MLR or ANN model, and provides a clue that the method proposed in this study may be useful in computeraided design of new polymethacrylates with desired T g .

show abstract

Section: Methodsmentioning

confidence: 99%

Modeling and Predicting the Glass Transition Temperature of Polymethacrylates Based on Quantum Chemical Descriptors by Using Hybrid PSO‐SVR

Pei

Cai

Zhu

et al. 2012

Macro Theory & Simulations

View full text Add to dashboard Cite

show abstract

“…Support vector machine (SVM) is a statistical learning approach based on structural risk minimization principle, which was proposed and developed by Vapnik and co-worker (1995). It has been successfully used for classification or regression in real applications (Cai et al, 2003a, b;Wen et al, 2009;Wang et al, 2011). The basic idea of SVR is to map the x into a higher-dimensional feature space F via a nonlinear mapping Φ(x), and then to perform a linear regression in this space.…”

Section: Theory Of Support Vector Regressionmentioning

confidence: 99%

Process analysis of electroless copper plating for AB5-type hydrogen storage alloy using support vector regression

Zhao¹

2012

Int. J. Phys. Sci.

View full text Add to dashboard Cite

Surface modification is an effective method to improve the electrochemical property of hydrogen storage alloy. In order to investigate the influence of process factors of electroless copper (Cu) plating for AB 5 -type hydrogen storage alloy on Cu coating mass, a novel modeling approach, support vector regression (SVR) combined with particle swarm optimization (PSO), was proposed to construct a mathematical model for prediction of the mass changes of Cu coating over the AB 5 hydrogen storage alloy surface based on three factors, including temperature, pH value and Ni 2+ concentration. The modeling accuracy and reliability of the created SVR model are validated through leave-one-out cross validation (LOOCV), and compared with those of a second-order polynomial model. The results show that the predicted errors by SVR-LOOCV models are all smaller than those achieved by the secondorder polynomial model. The SVR model is further applied to predict the process parameters for the maximum Cu coating mass. These studies suggest that SVR can be used as an effective methodology to assist the design of experiment, and is helpful to precisely control the coating mass via fine adjustment of the process parameters.

show abstract

“…At present, it has become a focus in machine learning research and is extensively employed in a wide range of real-word problems. [23][24][25][26][27][28][29][30][31] When SVM is applied to regression by introduction of an alternative loss function, it is termed as SVR. Suppose a sample is described by (x, y), where x represents the independent variable and y the dependent response.…”

Section: Support Vector Regressionmentioning

confidence: 99%

Prediction of the Glass Transition Temperatures of Styrenic Copolymers by Using Support Vector Regression Combined with Particle Swarm Optimization

Pei

Cai

Tang

et al. 2011

Journal of Macromolecular Science, Part B

View full text Add to dashboard Cite

Based on three quantum chemical descriptors (the average polarizability of a molecule (α), the most positive net atomic charge on hydrogen atoms in a molecule (q + ) and the heat capacity at constant volume (C v ) derived from the monomers using the density functional theory (DFT), the support vector regression (SVR) approach combined with particle swarm optimization (PSO), is proposed to establish a model for prediction of the glass transition temperature (T g ) of random copolymers including poly(styrene-co-acrylamide) (SAAM), poly(styrene-co-acrylic acid) (SAA), poly(styrene-co-acrylonitrile) (SAN), poly(styrene-co-butyl acrylate) (SBA), poly(styrene-co-methyl acrylate) (SMA), poly(styrene-co-ethyl acrylate) (SEA), and poly(acrylonitrile-co-methyl acrylate) (ANMA). The mean absolute error (MAE = 1.6 K), mean absolute percentage error (MAPE = 0.45%), and correlation coefficient (R 2 = 0.9978) calculated by SVR model are superior to those (MAE = 5.47 K, MAPE = 1.51%, and R 2 = 0.9829) achieved by a quantitative structure-property relationship (QSPR)/multivariate linear regression (MLR) model for the identical training set, whereas the MAE = 3.03 K, MAPE = 0.90%, and R 2 = 0.9952 calculated by SVR also outperform those (MAE = 5.38 K, MAPE = 1.61%, and R 2 = 0.9778) achieved by the QSPR/MLR model for the identical validation set, respectively. The prediction results strongly support that the modeling and generalization ability of the SVR model consistently surpasses that of the QSPR/MLR model by applying identical training and validation samples. It is demonstrated that the established SVR model is more suitable to be used for prediction of the T g values for unknown polymers possessing similar structure than the conventional MLR model. It is also shown that the hybrid PSO-SVR approach is a promising and practical methodology to predict the glass transition temperature of styrenic copolymers.

show abstract

Prediction of thermal conductivity of polymer-based composites by using support vector regression

Cited by 11 publications

References 29 publications

Modeling and Predicting the Glass Transition Temperature of Polymethacrylates Based on Quantum Chemical Descriptors by Using Hybrid PSO‐SVR

Modeling and Predicting the Glass Transition Temperature of Polymethacrylates Based on Quantum Chemical Descriptors by Using Hybrid PSO‐SVR

Process analysis of electroless copper plating for AB5-type hydrogen storage alloy using support vector regression

Prediction of the Glass Transition Temperatures of Styrenic Copolymers by Using Support Vector Regression Combined with Particle Swarm Optimization

Contact Info

Product

Resources

About