Impact of aggregate gradation and asphalt-aggregate ratio on pavement performance during construction using back propagation neural network

Abstract

This paper evaluates the influences of variations in asphalt mixture parameters during construction on the variations of pavement performance using back-propagation (BP) neural networks . The variations of gradation ( V G ) and asphalt-aggregate ratio ( V R a ) were assessed through a variability analysis . The influences of V G and V R a propagation were analyzed via BP neural networks and a sensitivity analysis. A reliability assessment was conducted to evaluate the joint effects of V G and V R a . Results illustrate that the V G and V R a during transportation are more severe than those during other processes. BP neural networks can precisely and robustly trace the influences of the V G and V R a . Pavement performance exhibits greater sensitivity to V R a and V G at sieve sizes of 0.075 mm and 2.36 mm. The joint effects of V G and V R a significantly degrade permanent deformation more than fatigue life . High-quality paving effectively mitigates the negative impacts of segregation during transportation. • Variations of gradation ( V G ) and asphalt-aggregate ratio ( V R a ) are more severe during transportation. • BP neural networks effectively trace the fluences of V G and V R a on pavement performance. • The joint effects of V G and V R a degrade permanent deformation more than fatigue life. • High-quality paving mitigates segregation impacts during transportation.