刚性基层沥青路面参数动态反演

为了对刚性基层沥青路面结构的模量反演提供参考，采用谱单元法作为正分析方法，利用实测弯沉时程曲线和理论计算结果构造优化目标函数，通过数值优化方法对沥青路面力学参数进行动态反演，并与传统静态方法反演的结果进行了比较。结果表明：传统静态方法反演得到各结构层的弹性模量，动态方法反演能得到沥青层动态模量主曲线和相位角主曲线以及其他结构层的弹性模量；动态模量和相位角主曲线组成完整的黏弹性参数，可以更全面地描述沥青混合料的力学特性与频率的相关性；动态反演方法采用各传感器处弯沉时程曲线作为优化过程中的约束条件，静态反演方法中仅采用弯沉曲线峰值作为优化约束条件，约束力度远小于动态反演方法，导致静态反演结果的变异性远大于动态反演结果，尤其是基层和底基层，静态方法反演结果的变异系数可达动态反演结果的2倍以上，而且反演得到的各结构层的弹性模量处于材料典型取值范围内的百分比明显小于动态反演结果。静态反演方法易引起"模量窜层"现象，造成反演结果失真，无法客观地对路面结构层进行质量评定；动态反演方法能利用实测弯沉时程曲线的全部信息，保证反演结果收敛于真实值，有效避免"模量窜层"现象，为路面质量评定提供有效途径。

Dynamic Backcalculation for Parameters of Asphalt Pavement with Rigid Base

The objective of this study is to provide reference for the backcalculation of asphalt pavements with a rigid base. Aiming at asphalt pavements with a rigid base, the spectral element method was used as the forward analysis engine, and then the optimized objective function was constructed using the measured and computed deflection time history. The layer parameters of the asphalt pavement were backcalculated using a numerical optimization method. Accordingly, the backcalculated results were compared with those obtained using the traditional static method. The results indicate that the backcalculated results from the traditional static method contain the elastic modulus of each layer. The dynamic modulus master curve and phase angle master curve of asphalt layers as well as the elastic moduli of other layers can be backcalculated from the dynamic method. The dynamic modulus master curve and phase angle master curve compose the complete viscoelastic parameters, which can completely represent the relationship between the viscoelastic property of the asphalt concrete and loading frequency. The dynamic backcalculation method uses the entire measured deflection time history at various sensor locations as the constraints in the optimization process. In comparison, the static method uses only the peak deflections as the constraints, so that its optimized constrain power is significantly lower than that of the dynamic method. The coefficient of variance (COV) of the backcalculated results from the static method is also larger than that from the dynamic method. Specifically for the base and subbase layers, the COV of the backcalculated results from the static method is more than twice that from the dynamic method. Hence, the percentage of the backcalculated result in the typical reasonable range is significantly smaller than those from dynamic method. The static backcalculation method is vulnerable to cause the "modulus-transferring" phenomenon, which will lead to distortion of the backcalculated result and prevent the objective assessment of the layer condition of the asphalt pavement. The dynamic backcalculation method could utilize the full information contained in the measured deflection time history, which could lead optimized results to actual values, avoid the "modulus-transferring" phenomenon and provide more effective approach for pavement quality assessment.