BP神经网络的隧道沉降预测虚拟仿真实验教学  

Research on Virtual Simulation Experimental Teaching of Tunnel Subsidence Prediction Based on BP Neural Network

作　　者：丁杨 韩震 张小龙 张鸿乾 周双喜 饶军[5] DING Yang;HAN Zhen;ZHANG Xiaolong;ZHANG Hongqian;ZHOU Shuangxi;RAO Jun(Department of Civil Engineering,Hangzhou City University,Hangzhou 310015,China;Nanjing Metro Operation Co.,Ltd.,Nanjing 210012,China;College of Civil Engineering and Architecture,Zhejiang University,Hangzhou 310058,China;School of Civil Engineering and Engineering Management,Guangzhou Maritime University,Guangzhou 510725,China;School of Civil Engineering and Architecture,East China Jiaotong University,Nanchang 330013,China)

机构地区：[1]浙大城市学院土木工程系,杭州310015 [2]南京地铁运营有限责任公司,南京210012 [3]浙江大学建筑工程学院,杭州310058 [4]广州航海学院土木与工程管理学院,广州510725 [5]华东交通大学土木与建筑学院,南昌330013

出　　处：《实验室研究与探索》2024年第1期78-81,138,共5页Research and Exploration In Laboratory

基　　金：国家自然科学基金(52163034);浙江省教育科学规划课题(2023SCG222);浙江省教育厅科研项目(Y202248682)。

摘　　要：随着人工智能技术的快速发展,传统土木工程课程迫切需要转型,其中虚拟仿真实验教学是提高高校智能建造教学质量的重要环节。借助Matlab仿真软件,搭建多层逆向传播(BP)神经网络预测模型,开展地铁盾构施工过程中地表沉降预测,分析不同输入量及隐含层数量等因素对预测性能的影响。仿真结果表明:对单隐含层神经网络而言,建议输入量为1,隐含层数量在1~6之间;对于双隐含层神经网络,建议输入量为5,第1、2层隐含层数量在1~3之间;所建单/多层BP神经网络能很好地预测沉降变化趋势,为智能建造课程的虚拟仿真实验教学提供借鉴。With the rapid development of artificial intelligence technology,traditional civil engineering courses urgently need to be transformed,and virtual simulation experimental teaching is an important part of improving the teaching quality of intelligent construction in universities.With the help of Matlab simulation software,a multilayer BP neural network prediction model is built to predict the ground settlement during the construction of subway shields,and analyze the influence of different input variables and the number of hidden layers on the prediction performance.The results show that for a single hidden layer neural network,it is recommended that if the input variable is 1,then the number of hidden layers should be 1 to 6;for a double hidden layer neural network,it is recommended that if the input variable is 5 then the number of hidden layers in the first and second layers should be 1 to 3.The established single/multi-layer BP neural network can well predict the trend of settlement changes,and can provide a reference for virtual simulation experimental teaching in intelligent construction courses.

关 键 词：智能建造 虚拟仿真 教学改革 地铁盾构 地表沉降 BP神经网络 

分 类 号：U231.3[交通运输工程—道路与铁道工程] G642[文化科学—高等教育学]