Viscoelastic modulus backcalculation of flexible pavements from traffic speed deflectometer measurements

Traffic Speed Deflectometer (TSD) has become a widely used tool for non-destructive evaluation of flexible pavements due to high efficiency in measuring surface deflections under moving loads. However, backcalculation of viscoelastic modulus from TSD data remains challenging due to time- and temperature-dependent behavior of asphalt materials. This study proposes an artificial neural network (ANN)-based framework to backcalculate viscoelastic modulus of asphalt layers from TSD-measured deflection slopes. A semi-analytical finite element method (SAFEM) was used to generate a comprehensive synthetic database covering various pavement structures, temperatures, loading speeds, and material properties. The viscoelastic modulus was represented using a sigmoidal function with four parameters, along with a time-decaying damage model to account for material degradation. The results indicate that TSD responses are primarily influenced by asphalt layer thickness based on normalized sensitivity indices. The ANN model demonstrates high prediction accuracy and robustness, as compared to other shallow machine learning models. Field validation using TSD measurements from four pavement sections further confirms that the backcalculated modulus is consistent with laboratory measured results. Overall, the proposed approach provides an efficient and reliable tool for viscoelastic modulus backcalculation from TSD data, with potential application for project-level pavement condition evaluation and rehabilitation design.