Prediction of asphalt complex viscosity by artificial neural network based on Fourier transform infrared spectroscopy

A predictive model was established by artificial neural network (ANN), and the complex viscosity * of asphalt was estimated based on Fourier Transform infrared (FTIR) spectroscopy. By linearizing the logarithmic model of complex viscosity, constants and were obtained as independent variables in ANN, and spectral data were taken as dependent variables. Comparing the measured values with the predicted values, it is found that the predicted and actual values of the *-T curve are consistent. It provided an innovative method basis for rapid prediction of asphalt viscosity by FTIR in engineering application.