基于XGBoost国省干线沥青路面使用性能预测研究
Research on Service Performance Prediction of Asphalt Pavement Based on XGboost National and Provincial Trunk Line
侯美晴 1粟海涛 2杨洋 1亓祥宇 2刘名扬1
作者信息
- 1. 昆明理工大学建筑工程学院防灾减灾重点实验室 昆明 650000
- 2. 云南省公路科学技术研究院 昆明 650000
- 折叠
摘要
文中提出了一种基于XGBoost(extreme gradient boost)算法的国省干线沥青路面损坏状况指数(pavement condition index,PCI)预测模型,并与传统回归模型预测结果进行对比.结果表明:通过对比分析不同模型的3个定量评价指标(均方根误差、平均绝对误差和拟合系数)以及最终预测结果的折线图,证明采用考虑多因素条件的XGBoost算法PCI预测模型的预测精度要优于传统回归预测模型,验证了该模型的有效性和优越性.
Abstract
A prediction model of pavement condition index,PCI)of national and provincial trunk lines based on XGBoost(extreme gradient boost)algorithm was proposed.Compared with the traditional regression model,the results show that by comparing and analyzing three quantitative evaluation inde-xes(root mean square error,average absolute error and fitting coefficient)of different models and the line chart of the final prediction results,it is proved that the prediction accuracy of PCI prediction model with XGBoost algorithm considering multi-factor conditions is better than that of traditional re-gression prediction model,and the effectiveness and superiority of this model are verified.
关键词
道路工程/沥青路面/路面性能预测/路面状况指数(PCI)/机器学习XGBoostKey words
road engineering/asphalt pavement/pavement performance prediction/pavement condi-tion index(PCI)/machine learning/XGBoost引用本文复制引用
基金项目
国家自然科学基金项目(11962009)
云南省交通运输厅科技创新及示范项目(云交科教便[2021]86号-三)
出版年
2024
NETL
NSTL
万方数据