Back-Calculation of Nonlinear Properties of Subgrade Using a New In Situ Nondestructive Testing Approach Based on Loading Sequence

This study aims to develop a new in situ nondestructive testing (NDT) method to determine nonlinear parameters of subgrade soils and to establish a method to calculate the subgrade design modulus. The new method collects nonlinear characteristics of the subgrade by in situ test with various combinations of dynamic stress and static stress and back-calculates subgrade nonlinear parameters by machine learning algorithm. The highlight of this study is to solve the problem that the results of different test methods are not unique, and the stress-dependence characteristics of the subgrade are fully considered. The main work is as follows: (1) develop a new resilient modulus prediction model considering dynamic and static effects; (2) propose a novel in situ NDT method based on loading sequences; (3) establish a fuzzy back-propagation neural network (BPNN) algorithm to compute subgrade nonlinear parameters; and (4) propose a calculation method of subgrade design modulus based on the principle of deflection equivalence and establish an empirical equation of subgrade design modulus. The results showed that the contact stress has an enhancing effect on resilient modulus of subgrade soils. The new in situ NDT method has been proved feasible in theory and can accurately obtain the nonlinear characteristics of the subgrade.