The Study on Backcalculation of Pavement Moduli with Genetic Algorithms

Most existing backcalculation programs using a numerical analysis method need both the extent and initial values of elastic moduli. They, therefore, have defect such as the accuracy of solution can become different because they react very sensitively in those values. To solve these problems, the GABack program, which applied genetic algorithms to backcalculation process, was developed in this study. This program can backcalculate global optimum solution regardless of initial values of elastic moduli. The result of GABack program was analysed with that of common backcalculation programs by using surface deflection data of asphalt concrete pavement measured by a FWD equipment. From this test, the results of GAback program was reasonable because results of GABack and MICHBACK program expressed comparatively smaller % error value than that of other programs. In this study, the generation No.100 was proposed as a second terminating condition because almost all results of convergence tendency analysis using a mean object function value of parent pool were converged between No. 20 and No. 80.