Research On Pavement Roughness Detection Method Based On Multi-Sensor Fusion

In recent years,China’s infrastructure construction has been developing rapidly,and in the field of highway transportation,as the mileage of highways of all grades continues to increase,the road maintenance and management work has become increasingly heavy.Pavement roughness is an important index of pavement performance,traditional pavement roughness inspection methods are less efficient and cannot be measured in a wide range,while modern measurement means such as laser leveling meter are expensive and have a single application.To address this problem,this paper proposes a pavement roughness detection method based on vehicle-mounted multi-sensor fusion.Based on the vehicle-mounted approach,two core sensors,3D LIDAR and vibration sensors,are used in this study,and a hardware design scheme as well as a synchronous acquisition scheme are proposed.In terms of hardware,the overall design and modification of the detection vehicle is carried out,and the hardware design scheme of each module is given.In order to achieve synchronous data acquisition,this paper realizes real-time data transmission between multiple sensors based on the real-time position information and time information provided by the position sensors through RS485 protocol serial communication,and also gets the actual hind measurement distance of the LIDAR through the installation attitude of the LIDAR and vibration sensors,so as to achieve the synchronous acquisition between multiple sensors in time and space.In this study,the line scan LIDAR is used to collect the point cloud data of multiple cross-sections on the road at equal intervals during the moving process,and these point cloud data can be obtained after point cloud pre-processing,coordinate conversion and inter-frame matching,from which the longitudinal section sequence is extracted;in order to reduce the influence of car vibration on the LIDAR,the vibration signal obtained from the vibration sensor is used to carry out wavelet transform and select the optimal In order to reduce the influence of vehicle vibration on LIDAR,the vibration signal obtained from the vibration sensor is wavelet transformed,and the optimal number of decomposition layers is selected and then decomposed and reconstructed to obtain the effective vehicle body vibration displacement compensation value;finally,the fusion model obtained from the longitudinal section sequence after vibration displacement compensation is substituted into the international roughness index model to obtain the international roughness index IRI.The experimental scheme was designed to verify the reliability of this research method.Firstly,the calibration of the equipment was completed,and then the repeatability test was conducted under the same conditions,and the obtained coefficient of variation was less than 5%,which initially proved the reliability of the method in this paper.After that,a section of rural road in Fuxin Mongol Autonomous County was selected for example analysis,and the road driving quality index RQI was introduced to evaluate the road.Under normal driving conditions,the pavement roughness data was collected by this method,and the correlation test analysis was conducted by using the roughness profile meter at the same time.The roughness data obtained by the two methods have obvious correlation,which makes this method verified again.This paper has 42 figures,11 tables and 58 references.