沥青路面车辙形貌时空演化行为与层位损伤关系研究

Currently the one-dimensional depth indicators for asphalt pavement rut are difficult to fully and accurately describe the morphology features of rut, to establish the relationship between the index and the causes of the rut, to reflect the evolution behavior under real-world conditions and other issues. The research develops a three-dimensional morphology simulation equipment on the basis of the on-site investigation of the rut morphology, the laser point cloud data of simulated rut morphology is acquired by the use of new 3D laser detection technology, the adaptive extraction algorithm of rut’s multi-dimensional indicators is presented; Continuously changing rut three-dimensional model of typical rutted road in field is acquired through the outdoor test , based on the use of mathematical statistics, the evolution patterns of multi-dimensional indicators over time, the index changing characteristics in different stages of ruts development and the effects of load, temperature and other random factors on index are quantified from the perspective of time and space, the dynamic relation between multi-dimensional indicators and different layers of damage is established combined with drill core sampling data; Through the analysis and screening of the multi-dimensional indicators’ information content and independence, rut horizons damage probability recognizing model is established by the use of radial basis neural network. The research is of great significance to elucidate the rut evolution behavior and to achieve automatic and nondestructive horizon damage identification.

针对目前沥青路面车辙的一维深度指标难以全面、准确描述车辙形貌特征，难以建立指标与成因之间的相互关系，难以反映真实条件下的演化行为等问题。本研究通过现场车辙形貌特征调查，研发车辙三维形貌模拟设备，基于新兴的三维激光技术获取模拟车辙激光点云数据，建立车辙多维度指标自适应提取方法；获取现场典型车辙路段连续变化的多维度指标，从时间角度，将车辙沿纵向划分若干单元，研究各单元多维度指标变化规律，将荷载、温度变化曲线与指标变化曲线相匹配，分析对车辙形貌演化过程影响；从空间角度，对比不同车辙多维度指标，研究分布区间、分布密度等特性以划分演化阶段，对不同演化阶段芯样进行图像扫描，识别各层永久变形曲线，计算层位损伤比率，建立多维度指标与层位损伤关联；采用主成分分析，考虑指标信息量与独立性进行筛选，采用径向基神经网络建立车辙层位损伤识别模型。研究成果对阐明车辙演化行为、实现层位损伤自动、无损识别具有重要意义。

项目基于新兴的路面三维激光检测技术，针对沥青路面车辙病害开展了三维检测与重构方法研究，并在此基础上对车辙破坏层位无损识别进行初步探索。项目首先分析了检测车辆横向偏移和不同数据密度对多点激光车辙深度检测准确性影响规律，在实际检测过程中，车辆行驶应偏向最深辙槽和槽壁坡度大的方向以全面获取最深辙槽形态，并采用小10m间隔纵向数据以准确识别车辙；其次开发了室内横、纵向可调节车辙形貌模拟设备，用以验证激光检测准确性；再次，利用三维激光检测车获取了GPS和车辙横断面高程等多源数据，通过道路中心线坐标提取与插值、多源坐标系统空间定位、车辙横断面高程信息标定及规则格网模型重构等技术手段实现了考虑道路线形的车辙三维重构；提出了车辙深度、车辙宽度、车辙正负面积和车辙正负体积等全面描述车辙形貌的多维度特征参数计算方法，并基于Matlab软件开发计算程序；分析了实测车辙路段典型特征指标在横向、纵向、垂向上的分布区间、分布密度、增长量等变化规律；最后，基于现场车辙横断面检测数据结合钻芯取样结果，构建了集成随机森林、逻辑回归、朴素贝叶斯的多分类器识别分类模型。研究为沥青路面车辙破坏层位的无损识别提供新的思路和理论依据。