Comparative analysis of empirical decomposition algorithms in predicting tire-pavement friction from asphalt surface textures: a Hilbert–Huang transform (HHT) analysis

The frictional properties of pavement are contingent on its surface texture, which consists of multiple scales that each contribute differently to friction generation at the tire-surface interface. Consequently, this study applied three adaptive decomposition algorithms of the Hilbert–Huang Transformation (HHT) to extract essential texture parameters from surface profiles, aiming to improve understanding of how pavement texture impacts friction properties. The decomposition algorithms used are Empirical Mode Decomposition (EMD), Ensemble EMD (EEMD) and Complete EEMD with Adaptive Noise (CEEMDAN). A Circular Track Meter (CTMeter) was used to measure the surface texture profiles of twelve asphalt mixture slabs with high macrotexture, while friction properties were assessed using a dynamic friction tester (DFT). Additionally, sixteen texture parameters related to height, spacing, and shape were computed to characterize the surface texture of both the original profile and the derived intrinsic mode functions (IMFs). The statistical analysis revealed that while the third IMF (IMF3) obtained by the EEMD exhibited the strongest correlation with the original profile, reaching up to 0.69, the IMFs obtained by EMD demonstrated the highest correlation with the DFT results. Under different polishing conditions and DFT speeds, the friction-texture correlation varied significantly, with the highest correlation observed at the peak values of DFT results and DFT values at 20 km/h (DFT20). Among the texture indicators, in addition to the mean profile depth (MPD), the root mean square (Rq), root mean square wavelength (λq), and two points slope variance (SV2) were recommended for characterizing both the original texture and IMFs profiles in the correlation analysis of the friction-texture relationship.