Prediction model of transverse cracks of asphalt pavement based on XGBoost algorithm
Transverse cracks are the main form of distress in asphalt pavement,and the accuracy of their pre-diction directly affects the reliability of pavement structure design.In order to accurately predict the damage of transverse cracks in asphalt pavement during use,a transverse crack prediction model based on the XGBoost algorithm is proposed.The model's performance can be enhanced by optimizing its hyperparameters with the TPE-BO method.Compared with RF and CatBoost models,the proposed model has higher prediction accuracy.In addition,the study evaluates the importance of features through correlation analysis and the SHAP method.The results show that the model achieves the best performance when the number of input variables is reduced by 4,reducing the cost and difficulty of data collection,which is of great significance for improving the econom-ic benefits of highway maintenance.
国家科技期刊平台
NSTL
万方数据
