Xác định mô đun đàn hồi của đất nền sử dụng mô hình bagging-based ensemble

Abstract

The resilient modulus (Mr) of pavement subgrade soils has been characterized to describe the structural response of pavements. It has been used as an essential property both for evaluating a reliable design and the performance of pavements. Accurate estimating the resilient modulus allows to early determine the response of subgrade soil under traffic loading. This paper proposes an ensemble machine learning technique, namely bagging to estimate the resilient modulus of subgrade soils. Individual and bagging-based ensemble learning classifiers are constructed from four different base learners, including artificial neural network (ANN), support vector regression (SVR), classification and regression tree (CART), and linear regression (LR). Prediction performances of bagging-based models are compared with those of individual ones via error rates. Analytical results indicated that the bagging-based ANN model outperformed other machine learning techniques. This paper, therefore, potentially provides a quick and reliable solution in early determining the resilient modulus of subgrade soils.