More accurate modulus back-calculation by reducing noise information from in situ-measured asphalt pavement deflection basin using regression model

摘要

The noise in in-situ measured asphalt pavement deflection basin is the primary cause for the difference between the theoretical deflection basin in back-calculating software and measured basin, which affects the accuracy of the modulus back-calculation. The influence of noise, including the pavement surface condition and internal anomalies, on the deflection basin was studied. The optimum mathematical regression model for the theoretical deflection basin without interference was analysed using the probabilistic density method, which was then applied to remove the noise information of the in-situ measured deflection basin. The back-calculated moduli (with and without deflection basin correction) were compared, and splitting moduli from coring samples of the surface asphalt layer were additionally compared for partial validation. The results show that the quintic polynomial is effective at eliminating the noise information of deflection basin. The corrected back-calculation moduli are reasonable and close to the splitting moduli, and its data variation is smaller than that of the uncorrected back-calculation moduli. (C) 2017 Elsevier Ltd. All rights reserved.