Temperature correction for traffic speed deflectometer measurements on flexible pavement using ANN models

This study aims to develop a robust temperature correction method for TSD measurements on flexile pavements using Artificial Neural Network (ANN). Semi-analytical finite element method (SAFEM) was used to simulate TSD with field validation. A database of TSD measurements was developed accounting for influencing factors of viscoelastic material properties of asphalt layer, pavement structures, speeds, and temperatures. Regression analysis indicated that TSD-measured deflection slopes are more sensitive to temperature than speed, with 20 km/h increase in speed approximately equivalent to 1°C decrease in temperature. ANN models are developed for temperature correction of TSD-measured deflection slopes at each offset using the inputs of raw measurements and mid-depth temperature of asphalt layer. The model accuracy is high with R-square values over 0.95 except the farthest sensor location at 1.5 m. The ANN model for predicting temperature correction factors (TCFs) of surface deflections was further developed and validated by TSD measurements in the field.