Journal of
Civil Engineering and Construction Technology

  • Abbreviation: J. Civ. Eng. Constr. Technol.
  • Language: English
  • ISSN: 2141-2634
  • DOI: 10.5897/JCECT
  • Start Year: 2010
  • Published Articles: 140

Full Length Research Paper

Predicting complex shear modulus using artificial neural networks

Abdulhaq Hadi Abedali
  • Abdulhaq Hadi Abedali
  • School of Engineering, University of Liverpool, Liverpool, United Kingdom
  • Google Scholar

  •  Received: 01 February 2015
  •  Accepted: 30 March 2015
  •  Published: 30 April 2015


Developing a predictive model for complex shear modulus of the asphalt binder is a complex technique due to several factors that affect the model’s estimating capability, such as rheological properties and test conditions. Several models were developed in this regard; some of these are linear regression models and relate to rheological properties of asphalt binder. A computational model based on artificial neural networks (ANNs) was used for developing models to predict complex shear modulus of the asphalt binder tested in DSR. In this study, two G* prediction models were developed and implemented based on experimental observations: Artificial Neural Network (ANN) model and Multi Linear Regression (MLR) model. A Feed Forward Neural Network (FFNN) model was applied to predict the G*. In order to evaluate the incomes of the two models, statistical parameters were used to make the comparison between them. These parameters include the correlation coefficient (R) and Average Percentage of Error (APE). The data set that has been used in this study includes temperature, frequency, phase angle, dynamic shear viscosity, shear stress and strain. The models were trained with 75% of experimental data. The ANN and MLR approaches were applied to the data to derive the weights and the regression coefficients respectively. The performance of the model was evaluated by using the remaining 25% of the data. By comparing the R2 of the models, the study reveals that the ANN model can be used as an appropriate forecasting tool to predict the G*, as it out-performs the MLR model.


Key words: Complex shear modulus, artificial neural network analysis, dynamic shear rheometer, multi-linear regression model.