Constitutive Model Prediction for Grouted Sleeve Connections of Highway Bridges Using the Machine Learning Method

The grouted sleeve connection is a key component in the precast highway bridge construction and requires further investigation of the theoretical design. This paper develops a novel method to obtain the constitutive model of the grouted sleeve connections with the machine learning (ML) method. The experimental test is conducted to study the mechanical properties of the grouted sleeve connections. Then, an ML method is developed to deal with the significance of design parameters and predict the constitutive model. Finally, the accuracy of the prediction method is verified by the test data. The results show 8.0% and 3.9% maximum errors in the prediction of strain and stress, respectively. The algorithm proposed in this paper performs well in predicting constitutive models. This study provides new perspectives on theoretical research in transportation construction.