基于介电特性的沥青混合料密度预估优化模型

To improve the accuracy of the traditional density prediction model for asphalt pavement, a parameter optimization of the asphalt mixture density prediction model was studied. Based on the dielectric mixing model of the asphalt mixture, the general formula of the asphalt mixture density prediction theory model was derived by introducing the scatterer influence coefficient v and shape factor u. The relative permittivities of different asphalt mixture samples and their components (asphalt and aggregate) were measured by Percometer tester. Based on the Levenberg-marquardt method and the Universal Global Optimization algorithm, the minimum error between the density estimated by the general formula and the value measured by the surface dry method was solved, which gavethe v and u optimal solutions of the influence coefficient and shape factor optimization (ISO) model.The reliability of the ISO model was tested using the Shapiro-wilk (SW) method. Finally, the superiority ofthe ISO model was verified by comparing different density prediction models. The research results show that the v and u values of ISO model are 5.1 and -4.5, respectively. It is effective and reliable to use ISO model to estimate the density of asphalt mixture. The smaller the nominal maximum particle size of the dense-graded asphalt mixture is, the higher the goodness of fit between the density predicted by the ISO model and the measured value, and the average relative erroris on the contrary. Compared with the Rayleigh and the Al-qadi, Lahouarand Leng(ALL) models, the ISO model offers the best density prediction, which can significantly improve the density prediction accuracy of asphalt concrete with large porosity. This study provides a new method for more accurate prediction of asphalt pavement density.