A novel characterisation method for asphalt pavement structural performance assessment based on vehicle vibration data fusion

Abstract

ABSTRACT Rapid structural performance assessment is essential to ensure asphalt pavement condition and efficient road network operation. However, extracting deeper pavement structural performance information based on vehicle vibration remains a major challenge. This study proposes a vibration data fusion method to realise the task of accurate structural performance characterisation. A 2D reconstruction algorithm transforms raw vibration signals into vibration images. Based on reconstruction images generated from different sensors and sensing directions, a pixel-level fusion framework and a feature-level fusion framework were put to improve the characterization accuracy. The study extracted two effective image texture features to characterise asphalt pavement’s structural strength. Fitting with deflection data, the average goodness-of-fit of optimised indicators combination under a feature-level fusion framework can reach 0.827. It is 32.5% and 15.2% higher than two unfused single-index accuracy. The data fusion framework shows the practicality in evaluating pavement structural performance rapidly and accurately through the vehicle vibration response.