|
In the actual production on the Mix than design , in the case of gradation, aggregate type , asphalt varieties identified in advance does not know its asphalt aggregate ratio , volume indicators and anti-rutting performance meets the requirements , such as mixed the rut resistance test after the design is complete , the mixture ratio can not meet the requirements , but also on the level of asphalt mixture with composition readjust test , test time and effort . So it is necessary to establish a unified model to predict the asphalt mixture VMA and rut resistance , so as to guide the composition of the asphalt mixture gradation design . The central idea surrounding the subject of the following three aspects : First , the analysis of the VMA and influencing factors of resistance to rutting . The test results show , VMA by gradation , aggregate type is relatively large ; gradation, type of aggregate , asphalt type , aggregate ratio and porosity has an important influence against rutting . On this basis , explores the VMA 's influencing factors by multivariate linear regression VMA and sieve through rate of synthesis of bulk density , mineral aggregate , the the synthetic apparent density of the mineral aggregate relationship between mathematical formula better relevance and estimated VMA compared HOUDSON law ; each sieve through rate is derived by multiple linear regression based on the dynamic stability of the influencing factors , type asphalt , mineral aggregate synthetic bulk density , mineral aggregate synthetic apparent density , aggregate ratio gap the mathematical relationship between the rate , also made ??a good correlation ; viscoelastic asphalt mixture principle , according Burgers model the constitutive relation forecast dynamic stability . Second , we use the improved BP neural network processing . VMA and dynamic stability model training , generalization ability of the neural network , and forecast data using a variety of algorithms in MATLAB .
|