利用BP人工神经网络反算沥青路面基层弹性模量研究

基于层状弹性体系理论，利用BP人工神经网络预测沥青路面基层弹性模量。通过构建路表弯沉盆与沥青路面基层参数之间的数据库，建立了BP人工神经网络反算基层弹性模量预测模型。由理论弯沉盆作为已知输入参数进行反算时，基层弹性模量的反算值与理论值相对误差在8%左右；由实测弯沉盆作为已知输入参数进行反算时，计算弯沉盆与标准化实测弯沉盆的拟合均方差RMSE为4.49%。检验结果表明:基于层状弹性体系理论，建立的BP人工神经网络反算沥青路面基层弹性模量模型，可满足实践工程要求。

Back-calculation of Elastic Modulus of Asphalt Pavement Base by BP Artificial Neural Network

Based on the theory of layered elastic system, the elastic modulus of asphalt pavement base was predicted by use of BP artificial neural network. The paper constituted the database of surface deflection basin in relation to the parameters of asphalt pavement base, thus established a model predicting the elastic modulus back-calculated by BP artificial neural network. When the theoretical deflection basin was taken as the given input parameter for the back-calculation, the relative error between the back-calculated and theoretical values was about 8%; and when the actual deflection basin was taken, the fitted root mean square error (RMSE) between the calculated and standardized actual deflection basin was 4.49%. The check results show that, depending on the theory of layered elastic system, the model established herein for elastic modulus of asphalt pavement base back-calculated by BP artificial neural network, may meet the practical engineering requirement.