面向抗滑的路面多尺度特征识别方法研究

Pavement surface texture significantly contributes to skid resistance. Traditional approaches of characterizing pavement texture are limited in capability to reflect synthesized extension of the surface irregularities related to skid resistance. Motivated by the rough repetitiveness and self-similarity,this study attempts to develop a multifractal method to quantify the complexity and self-similarity of spatial arrangement of pavement surface texture and subsequently evaluate skid resistance. To achieve the objective, first, experimental data are employed and analyzed to justify the potential multifractal features of pavement surface profile. Second, the direct approach to compute multifractal spectra is explored. Specifically, the capability in explicitly distinguishing different types of pavement is revealed by multifractal spectra developed for the sampled surface textures. Third, multifractal parameters are developed to feature the pavement texture and explore the effects on skid resistance. For the purpose of evaluation, surface texture of pavement samples employed are measured via a laser and image analyzer-based device, and their skid resistance are acquired via the Portable Pendulum Tester and the Mu-meter MK6. Last, the model to evaluate pavement skid resistance using characteristic parameters of multifractal spectrum will be set up by multivariate statistics and artificial neural network. Further studies are needed to explore the potential application issues of this novel technique.

路面纹理是影响路面抗滑性能的关键因素之一，以往的研究仅从垂直方向、水平方向或单重分形层面刻画路面纹理整体的几何形貌，未能揭示出路面纹理所蕴含的整体相似性、局部差异性、复杂性等关键特征，因而也未能有效估计路面抗滑性能。本项目创新性的提出采用多重分形技术识别不同尺度下的路面纹理特征，试图找到能全面反映具有抗滑差异性的路面纹理之间的奇异谱及其快速求解方法，探寻最能表述纹理丰富信息的新参数集，并通过大量的实验分析与理论推导阐明新参数集对路面抗滑性能产生显著影响的物理意义；尝试采用多元统计和人工神经网络技术构建基于纹理多重分形特征向量的路面抗滑性能非线性最优估计模型及其校验方法，该模型可用于指导路面纹理施工，为铺制具有不同抗滑性能的路面提供理论支撑。

路面纹理是影响路面抗滑性能的本质性与关键性因素，继而影响胎-路之间的运行安全。项目主要研究了基于高精度激光测距仪的路面二维纹理采集技术、基于图像的路面三维纹理采集技术、基于多重分形理论的路面纹理多尺度特征表达方法、以及路面纹理多尺度特征与抗滑值的非线性关系等内容。研究结果表明：①激光测距仪与图像算法得到的路面纹理特征数据变化趋势一致，且相对误差较小，验证了基于图像分析技术采集路面三维纹理具有可行性。②路面局部纹理呈现出不同的奇异强度，具有多重分形特征；不同材料和级配组成的路面纹理具有不同的多重分形特征，若集料最大公称粒径越大、表面空隙越宽、结构越复杂，则分形谱宽度和最大、最小概率子集分形维数的差值越大，多重分形谱参数能够有效描述路面纹理的粗糙程度和尖锐程度；相同材料和级配组成的路面纹理具有相近的多重分形特征，多重分形谱形状及其参数能够稳定的识别这种聚类特征。③交互作用下的多重分形参数和垂直方向纹理参数与路面抗滑性能之间存在对数回归关系，其相关系数均超过0.8，拟合优度较高，当显著性水平为0.05时，t检验的回归系数均显著。本研究成果对于揭示路面纹理的抗滑机理、指导路面施工工艺、提升道路交通安全性具有科学意义。