Statistical analysis of aging micro-characteristics for asphalt binder AFM images based on neural network classification

Abstract

To further evaluate the effect of aging behavior on microstructure for asphalt binder, the atomic force microscope (AFM) images of asphalt binder were studied by neural network classification technology, and its aging micro-characteristics were analyzed statistically. Firstly, AFM equipment was used to test 90# virgin asphalt binder and styrene–butadiene–styrene (SBS) modified asphalt binder commonly used in engineering, and the typical micromorphology AFM images of asphalt binders were obtained. Subsequently, the asphalt binder AFM images were preliminatively analyzed, and the micro-characteristics of AFM images before and after aging were compared, including the “bee structure” characteristics, distribution law and surface structure roughness. Finally, the asphalt binder AFM images were further analyzed by neural network classification, and the change of “bee structure” was statistically calculated and analyzed, and its aging micro-characteristics were quantitatively evaluated. The results show that it is feasible for neural network classification to be applied to the statistical analysis of asphalt binder AFM images, and it can further describe and evaluate the aging micro-characteristics of asphalt binder. Aging behavior reduces the proportion of intermediate category areas in asphalt binder AFM images and increases the “bee structure” areas. The cumulative percentage curve of asphalt binder AFM images is smoother (each category is more uniform), indicating that the asphalt binder performance tended to be stable after aging. The results further enrich the description of asphalt binder AFM images, and provide a new way to explore the micro-morphology of asphalt binder.