Mechanical properties of frozen soils (e.g., triaxial compressive strength, and Young’s modulus, ) are important in tunnel, shaft, or open pit excavation projects. Although numerous attempts have been made to develop indirect methods to estimate unfrozen soils’ and values, this has not been done with frozen soils given the difficulty of preparing and conducting relevant laboratory tests. In this study, the accuracy of artificial neural network (ANN), adaptive neuro fuzzy inference system (ANFIS), and support vector machine (SVM) models, developed to predict and for frozen sandy soils, was compared. To the best of the authors’ knowledge, no study has predicted frozen soils’ and using these methods. Eighty-two poorly graded sandy soil samples from an urban subway borehole in Tabriz, Iran, were used to develop these models. It was found that temperature, confining pressure, strain rate, and yielding strain improved the accuracy of and prediction. Results indicate that SVM can successfully be used in predicting the and of frozen soils.
Resilient modulus ( ) plays the most critical role in the evaluation and design of flexible pavement foundations. is utilised as the principal parameter for representing stiffness and behaviour of flexible pavement foundation in experimental and semi-empirical approaches. To determine , cyclic triaxial compressive experiments under different confining pressures and deviatoric stresses are needed. However, such experiments are costly and time-consuming. In the present study, an extreme gradient boosting-based ( ) model is presented for predicting the resilient modulus of flexible pavement foundations. The model is optimised using four different optimisation methods (particle swarm optimisation ( ), social spider optimisation ( ), sine cosine algorithm ( ), and multi-verse optimisation ( )) and a database collected from previously published technical literature. The outcomes present that all developed designs have good workability in estimating the of flexible pavement foundation, but the models have the best prediction accuracy considering both training and testing datasets.