中文摘要：

在基于分形的沥青路面微观纹理特征描述中,如何选取合适的分形插值垂直比例因子是一个无公认选择标准的问题.文中提出一种适用于沥青路面的分形插值模型的改进算法,将遗传算法和序列二次规划方法结合应用于分形插值模型参数的寻优过程,并在自制沥青试样检测平台采集到的沥青路面微观纹理的基础上进行对比试验.结果表明,经组合优化后的分形插值方法在计算沥青路面复杂微观结构和小尺度特征描述等方面更具优势,它不仅克服了遗传算法中全局最优解求解精度低的问题,且使传统最优化方法的最优解的局限性得到了改善,进一步改进了分形插值结果的精度,实现了路面构造深度曲线的最佳拟合.

英文摘要：

Since there is no accurate method to calculate the fractal interpolation model of micro-texture on asphalt pavement, a hybrid optimization algorithm combined genetic algorithm （GA） and sequen- tial quadratic programming （SQP） is put forward in the research of vertical scaling factors in the fractal interpolation model. Additional, a high-precision laser range finder, which is constituted by laser sensor and chip STM32F417, was designed to measure the asphalt pavement texture for comparison. The comparison experiment showed that the fractal interpolation model calculated by hybrid optimization algorithm is better than the model optimized by GA in terms of fitness, approximation degree of interpolation curve and the accuracy of fractal dimension.