基于BP神经网络的土基回弹模量反算

基于三层BP神经网络和层状弹性理论，由路表变形反算土基回弹模量。利用层状弹性计算程序构建了土基回弹模量与路表弯沉数据库，并输入BP神经网络进行训练，建立了基于BP神经网络的土基回弹模量反算模型。理论和实测路表弯沉反算结果表明，该方法具有良好的识别能力和泛化能力，是一种实用而有效的方法，为进一步快速、有效地评定土基的承载能力提供了依据。

BP Artificial Neural Network-based Backcalculation of Subgrade Resilient Moduli

Based on a 3-layer BP artificial neural network and the layered elastic moduli was backcalculated by the surface deflections. The database of s moduli and surface deflections were established using layered elastic calculation program. Backcalculation model of subgrade resilient moduli was developed after the BP artificial neural network was trained by the established database. The results of backcal culation by theoretical and measured surface deflections indicate thatthe ability of this approach to re-recognition and generation is satisfactory and it is a simple and practical method. It provides evidences to rapidly and effective- ly evaluate the bearing capacity of subgrade.