Data fusion of ultraviolet-visible and mid-infrared spectra for rapid performance inspection of paving asphalt binders

The quality of paving binders is of great importance that affects the performance of asphalt pavements. In this study, a rapid performance characterization method was proposed for paving asphalt binders based on the multi-spectra data fusion and chemometrics, with the purpose of quality inspection. Firstly, a data set of paving binder samples considering multiple origins was collected, and various binder performance indicators were obtained with standard tests. Meanwhile, ultraviolet-visible (UV–Vis) and mid-infrared (MIR) spectra of binders were measured and fused at the data level; Secondly, the chemometric method, partial least squares regression (PLSR), was employed to extract information hidden in the fused spectra for the establishment of binder performance prediction models. The quality of most calibrated PLSR models was affirmed, and a good agreement between the measured and predicted performance indicators was observed. In particular, the rapid spectra measurements could save a lot of testing time, and the data fusion of UV–Vis and MIR spectra could provide more satisfactory outputs in most cases; Finally, the repeatability and reproductivity of the proposed method were validated, with standard deviations and coefficients of variation far below limits specified in the specification. The automated spectra measurements, appropriate spectra pretreatments, and the ease of sample preparation and instrument operation could account for this. Above all, the integrated spectra-chemometrics method proposed in this study had the potential of rapid binder performance inspection and quality assurance in practice, especially with more samples from different origins and production technologies collected to extend the existing dataset in the future.