Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
A Vibration Compensation Method for Pavement Depth Data Based on LightGBM with Bayesian Optimization
To salve the issue of local distortion and feature distortion caused by vibration during the data acquisition process of 3D pavement detection vehicles, this research proposed a Bayesian optimization-based LightGBM method for compensating for vibration in pavement depth data. Firstly, the least squares method is used to remove trend of the acceleration data and interpolate missing values. Then, the pavement depth data and acceleration signals are matched using timestamps to obtain the dataset. Finally, the vibration compensation for pavement depth data is achieved through Bayesian optimization-based LightGBM. By comparing Bayesian optimization, genetic algorithm, and grid search parameter optimization methods, the results demonstrate the superiority of the Bayesian optimization-based LightGBM, with an R2Score reaching 0.84. Additionally, the proposed method is validated with real- scene pavement depth data, further demonstrating its feasibility, effectiveness, and generalizability.
A Vibration Compensation Method for Pavement Depth Data Based on LightGBM with Bayesian Optimization
To salve the issue of local distortion and feature distortion caused by vibration during the data acquisition process of 3D pavement detection vehicles, this research proposed a Bayesian optimization-based LightGBM method for compensating for vibration in pavement depth data. Firstly, the least squares method is used to remove trend of the acceleration data and interpolate missing values. Then, the pavement depth data and acceleration signals are matched using timestamps to obtain the dataset. Finally, the vibration compensation for pavement depth data is achieved through Bayesian optimization-based LightGBM. By comparing Bayesian optimization, genetic algorithm, and grid search parameter optimization methods, the results demonstrate the superiority of the Bayesian optimization-based LightGBM, with an R2Score reaching 0.84. Additionally, the proposed method is validated with real- scene pavement depth data, further demonstrating its feasibility, effectiveness, and generalizability.
A Vibration Compensation Method for Pavement Depth Data Based on LightGBM with Bayesian Optimization
Kong, Xianglong (Autor:in) / Yang, Ming (Autor:in) / Yuan, Bo (Autor:in) / Ding, Jangang (Autor:in) / Li, Wei (Autor:in)
10.11.2023
3888631 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Prediction of the ROP based on GA-LightGBM and drilling data
| Taylor & Francis Verlag | 2025
| DOAJ | 2023
Pavement modelling based on measured in-depth deflection data
| British Library Conference Proceedings | 1993