神经网络的土钉轴力最大值预测模型  

Prediction Model of Maximum Axial Force of Soil Nail Based on Neural Network

作　　者：卓维松[1] 阙云[2] 王荣 黄瑞 ZHUO Wei Song;QUE Yun;WANG Rong;HUANG Rui(School of Civil Engineering,Fujian Chuanzheng Communications College,Fuzhou,Fujian 350007;School of Civil Engineering,Fuzhou University,Fuzhou,Fujian 350116;School of Civil Engineering and Architecture,Wuyi University,Wuyishan,Fujian 354300)

机构地区：[1]福建船政交通职业学院土木工程学院,福建福州350007 [2]福州大学土木工程学院,福建福州350116 [3]武夷学院土木工程与建筑学院,福建武夷山354300

出　　处：《武夷学院学报》2023年第9期69-74,共6页Journal of Wuyi University

基　　金：福建省自然科学基金项目(2020J01403)。

摘　　要：为提高土钉最大轴力的预测精度,收集大量土钉轴力的实测数据,建立考虑土钉墙参数、土钉参数、土参数和超载情况等因素的土钉轴力神经网络预测模型,并对该模型的预测结果进行评估。结果表明:预测模型的土钉轴力最大值与实测的土钉轴最大值吻合程度高,离群点较少;神经网络梯度稳步降低,参数Mu随着迭代次数稳定变化,最大值也未超过0.001,训练R值为0.981,验证R值为0.73,测试R值为0.82,整体R值为0.92;对比现有土钉轴力模型,预测模型的土钉轴力平均值更精确,模型因子变异性更小,且预测精度和土钉轴力最大值之间不存在相关性,因此适用性更广泛。In order to improve the prediction accuracy of maximum axial force of soil nail,a large number of measured data of soil nailing axial force were collected,and a neural network prediction model of soil nailing axial force was established considering soil nailing wall parameters,soil nailing parameters,soil parameters and overload condition,and the prediction results of the model were evaluated.The results show that the maximum value of soil nail axial force predicted by the model is in good agreement with the measured value,and there are few outliers.The gradient of neural network decreased steadily,the parameter Mu changed steadily with the number of epochs,and the maximum value did not exceed 0.001.The R value of train was 0.981,the R value of validation was 0.73,the R value of test was 0.82,and the overall R value was 0.92.Compared with existing soil nail axial force models,the average value of soil nail axial force of the prediction model is more accurate,and the variation of model factors is smaller.Moreover,there is no correlation between the prediction accuracy and the maximum value of soil nail axial force,so it is more widely applicable.

关 键 词：土钉墙 土钉轴力 神经网络 预测模型 变异性 

分 类 号：TV551[水利工程—水利水电工程]