modeling marshall test results of hot mix asphalt using nonlinear genetic programming techniques

The marshall test method is widely used for the design and control of hot mix asphalt (hma). the marshall and modified marshall mix design methods are most widely used in iran. determining marshall test results (marshall stability, flow, and marshall quotient (mq)) are time-consuming. therefore, using new and advanced methods to determine the results of marshall testing is essential. in this study, the genetic programming method based on artificial intelligence was used for the prediction of marshall test results. input variables in the genetic programming models use the volumetric properties of standard marshall specimens such as air voids, voids in mineral aggregate (vma), and voids filled with asphalt (vfa). also, multiple linear regression models were used as the base model to evaluate the models presented by the genetic programming method. the results indicated that the proposed methods are more efficient than the laboratory costly method and the performance of the genetic programming model is completely satisfactory in comparison to the base model and has been able to predict the results of marshall testing based on the input parameters. the gp models have a higher coefficient of determination and fewer errors than mlr models. the presented models will also help further researchers willing to perform similar studies, without carrying out destructive tests.