基于 BP 神经网络的降雨量对恒重式路肩挡土墙位移研究

摘要

This paper improves activation function in the structure of neural network into antisysmmetric function .The analysis to deformation of retaining wall by rainfall in practical engineering based on BP neu-ral network has been made .The use of monitoring data is to make training simulation ,and the actual a-mount of displacement is compared with theoretical exploration to provide design data and design reference for retaining wall design in engineering with similar geological conditions and hydrological conditions .%将传统神经网络结构中的神经网络激励函数改进为一反对称函数，运用BP神经网络对实际工程的降雨量对恒重式路肩挡土墙变形位移进行分析，利用监测数据进行训练模拟，并与实际的位移量进行对比，对相似工程地质条件、水文条件下路肩挡土墙设计提供了设计依据及设计参考。