柔性路面结构参数反算的人工神经元法

Artificial Neural Networks Method for backcalculation of flexible pavement structural parameters

利用传递矩阵的方法，计算出了柔性路面三层结构体系在轴对称荷载作用下，层间完全接触情况的表面弯沉值。利用计算数据训练了三层 BP网络，从而得到了较精确的权值和阀值，再利用所得到的权值和阀值对柔性路面结构的模量和厚度进行了反算，并将反算值与实际值进行了比较，得到了令人满意的结果。该法速度快、精度高，除了反算模量外，还可反算路面结构层的厚度。

In this paper,by means of transfer matrix,the superficial deflection of flexible pavement triple layered structure,in the axisymmetrical loading case in which layers are completely contacted,is calculated. With the data,three layers of BP networks are trained,so more accurate right value and threshold value are obtained. Again,with the right value and threshold value,the modulus and thickness of flexible pavement structure are backcalculated,and the calculated results which are compared with factual value are satisfactory. The accuracy and speed of this method are high. It can backcalculate both modulus and thickness of flexible pavement structure layers.