水泥砼路面基层刚度反演分析的EBP方法

利用人工神经网络方法，对水泥砼路面基层刚度的计算进行了研究。采用有限元模态分析技术，对弹性半无限空间体上的水泥砼路面板有、无基层刚度变化情况分别进行了模拟，得到了一系列路面板自振频率与基层平面位置的映射关系，将此作为EBP网络学习样本，为加快网络训练收敛速度，采用了学习率自适应调整策略。算例结果表明，EBP人工神经网络算法可用于确定路面板基层刚度变化与其实际位置之间的映射关系。

Back-analysis of Stiffness of Base Course in the Cement Concrete Pavement by EBP Method

The stiffiness calculation of base course in the cement concrete pavement is researced by artificial neural network method. A set of data about the self_vibration frequency changes are got,when the simulation by finite element method (FEM) modal analyses, for the cement concrete pavement on elastically half_infinite space foundation, is used in the cases of whether the base course stiffness is altered or not. The mappings of frequency changes and positions on the plane of base course are considered as training samples of the Error Back_Propagation (EBP) network, with the variable learning rate back_propagation algorithm are applied to convergence fast. The simulation results by EBP method show that the mappings of frequency changes and plane positions can be determined by artificial neural network algorithm.