市政技术2025，Vol.43Issue(5)：50-59,127,11.DOI:10.19922/j.1009-7767.2025.05.050
基于行车振动数据的非线性路面平整度检测模型研究
Research on Nonlinear Road Roughness Detection Model Based on Vehicle Vibration Data
张宇轩 1张金喜 2刘旎 3刘立宁 3蔡守利3
作者信息
- 1. 北京市政路桥管理养护集团有限公司,北京 100097||北京工业大学交通工程北京市重点实验室,北京 100124
- 2. 北京工业大学交通工程北京市重点实验室,北京 100124
- 3. 北京市政路桥管理养护集团有限公司,北京 100097
- 折叠
摘要
Abstract
In recent years,highway maintenance costs have continued to rise in China,making periodic pavement roughness detection crucial for optimizing maintenance processes.In order to reduce inspection costs and improve efficiency,a self-developed intelligent pavement roughness detection device was employed to collect vehicle vibra-tion data of expressways around Beijing.The data underwent verification,filtering for noise reduction,and feature indicator screening.Ultimately,multiple linear regression and XGBoost nonlinear regression model were established for pavement roughness prediction and analysis.The results demonstrate that the devices have great advantages of GPS positioning accuracy,data consistency and repeated testing reliability;Power Spectral Density(PSD)and average speed(Vave)were screened to be effective feature indicators;The XGBoost nonlinear regression model exhibited supe-rior predictive performance than multiple linear regression,with R2 test set of 0.610 7,indicating the feasibility of the detection plan based on vibration acceleration.However,since the experimental data primarily came from highways in"excellent"roughness conditions,the model's predictive capability was limited by the lack of data from poorly main-tained road sections.Future research should expand the scope of testing and include more diverse road conditions to refine the model and provide more efficient and cost-effective detection method for pavement maintenance management.关键词
IRI/智能检测/沥青路面/XGBoost模型/行车振动加速度Key words
IRI/intelligent detection/asphalt pavement/XGBoost model/vehicle vibration acceleration分类
交通运输引用本文复制引用
张宇轩,张金喜,刘旎,刘立宁,蔡守利..基于行车振动数据的非线性路面平整度检测模型研究[J].市政技术,2025,43(5):50-59,127,11.
