Crack Propagation and Intelligent Prediction in Asphalt Pavements Under Moving Loads

Asphalt pavements are prone to crack initiation and propagation under the interaction of moving loads and natural environmental conditions, significantly reducing their performance and lifespan. Guided by fracture mechanics theory, this study investigates the mechanisms and key influencing factors of crack propagation in asphalt pavements subjected to moving loads through an integrated approach combining finite element simulation with back propagation (BP) neural network-based prediction. A three-dimensional pavement model containing a longitudinal crack was developed in ABAQUS to analyze the evolution of stress intensity factors K_I and K_II at the crack tip. The influences of vehicle speed, load level, and structural parameters, including the thickness and elastic modulus of the surface, base, and sub-base layers, were examined. The results show that low-speed driving and overloading markedly increase the peak values of K_I and K_II, thereby accelerating crack propagation. A decrease in the thickness of the surface layer or an increase in its elastic modulus greatly raises stress intensity factors, while the influence of base and sub-base parameters is relatively limited. A decrease in surface layer thickness or an increase in its elastic modulus significantly elevates the stress intensity factors, whereas the effects of base and sub-base parameters are relatively minor. The developed BP neural network-based prediction model achieves accurate estimation of K_I and K_II, with average errors below 3%, thereby offering a practical and efficient tool for rapid assessment of pavement cracking resistance. These findings furnish a theoretical foundation for the optimized design of asphalt pavements and the development of maintenance strategies, while also establishing a basis for future research on crack propagation under multi-factor coupling conditions.