Road Roughness Assessment Based on Fusion of Connected-vehicles Data
The rapid development of intelligent connected vehicles(ICV)provides a new solution for monitoring infrastructure performance in a more convenient and scalable manner.To improve the efficiency and accuracy of the vehicle dynamic-response-based pavement roughness evaluation method,this study proposes a pavement profile inversion method based on an augmented Kalman filter using vehicle vibration response signals as the input.With the estimated pavement profile,the international roughness index(IRI)can then be calculated to evaluate the roughness of the pavement.Subsequently,a Gaussian process model was used to fit the estimated profile results from multiple vehicles to achieve a fusion resolution of the multi-vehicle results.With the help of the vehicle communication of intelligent connected vehicles,the fused results of previous vehicles can be incorporated as pseudo-measurements in an ego vehicle.Thus,the pavement profile estimation accuracy and robustness of the ego vehicle can be further improved.Numerical simulations and field tests are performed to validate the proposed method.The test results indicated that the proposed estimation method can provide good accuracy for IRI prediction.The correlation between predicted and measured IRI values was approximately 87％.The root mean square error of the predicted IRI value was approximately 0.26 m·km-1.Repeatability tests with multiple vehicles verified the accuracy and repeatability of this method.Moreover,the multi-vehicle result fusion and collaborative estimation strategy can not only reduce the error of the single-vehicle estimation result but also improve the stability of the single-vehicle estimation algorithm.Based on the algorithm validation,a prototype device for online pavement roughness evaluation was designed,and a framework for pavement roughness evaluation in an ICV environment was proposed and validated offline.Based on this framework,not only can the single-vehicle online estimation of pavement roughness be realized,but multivehicle estimation results can also be aggregated.The most practical appeal of this framework is that the fused results at the platform side can be further used in edge calculations to realize cloud-edge collaborative estimation and improve the accuracy and robustness of edge inversion.
NETL
NSTL
万方数据
