基于TimeGAN增强的CNN-LSTM模型在盾构掘进地表沉降中的预测研究

郁万浩; 刘陕南; 肖晓春

doi:10.3973/j.issn.2096-4498.2024.11.012

隧道建设 >

2024 , Vol. 44 >Issue 11: 2223 - 2232

DOI: https://doi.org/10.3973/j.issn.2096-4498.2024.11.012

研究与探索

基于TimeGAN增强的CNN-LSTM模型在盾构掘进地表沉降中的预测研究

郁万浩 ,
刘陕南 ,
肖晓春

展开

（1. 上海工程技术大学城市轨道交通学院，上海 201620； 2. 上海隧道工程有限公司，上海 200032 )

郁万浩（1999—），男，安徽明光人，上海工程技术大学交通运输专业在读硕士，研究方向为基于深度学习的盾构隧道施工沉降预测及控制。E-mail： M405121404@sues.edu.cn。*通信作者：刘陕南， E-mail: 10130004@sues.edu.cn。

网络出版日期: 2024-12-12

收起

Prediction of Shield Tunneling-Induced Surface Settlement Using Time Series Generative Adversrial Networks Enhanced Convolutional Neural Networks-Long Short-Term Memory Model

YU Wanhao ,
LIU Shannan ,
XIAO Xiaochun

Expand

(1. School of Urban Railway Transportation, Shanghai University of Engineering Science, Shanghai 201620, China; 2. Shanghai Tunnel Engineering Co., Ltd., Shanghai 200032, China)

Online published: 2024-12-12

Fold

摘要

为更准确地预测小数据量下盾构法施工造成的地表沉降，提出基于TimeGAN（time series generative adversarial networks，时间序列生成对抗网络）增强的CNN（convolutional neural networks，卷积神经网络）-LSTM（long short-term memory，长短期记忆网络）盾构掘进地表沉降预测模型，并依托上海北横通道新建工程Ⅱ标盾构施工项目验证该增强模型的性能。首先，选取300环的部分施工参数、地质参数、几何参数以及地表最大沉降，对比LSTM、CNN-LSTM与TimeGAN-CNN-LSTM的性能，证明CNN-LSTM对于盾构施工环境下多参数的预测效果明显优于LSTM，TimeGAN-CNN-LSTM增强模型优于CNN-LSTM；然后，通过更改训练集及测试集的大小，对不同数据集下TimeGAN-CNN-LSTM增强模型相较CNN-LSTM的预测效果进行研究。结果表明： TimeGAN-CNN-LSTM增强模型预测效果相较CNN-LSTM模型提升显著，且当训练集与测试集比值为4~8时，提升最为显著。

关键词： 盾构隧道; 地表沉降; 卷积神经网络; 长短期记忆网络; 时间序列生成对抗网络

本文引用格式

郁万浩 , 刘陕南 , 肖晓春 . 基于TimeGAN增强的CNN-LSTM模型在盾构掘进地表沉降中的预测研究[J]. 隧道建设, 2024 , 44(11) : 2223 -2232 . DOI: 10.3973/j.issn.2096-4498.2024.11.012

Abstract

The authors present a predictive model for shield tunneling-induced surface settlement using a convolutional neural networks (CNN)-long short-term memory(LSTM) framework enhanced by time series generative adversarial networks(TimeGAN). This model is designed to deliver accurate predictions even with limited data volumes. The model′s effectiveness is validated through its application to the section Ⅱ of the Beiheng tunnel project in Shanghai, China. Key parameters, including construction, geological, and geometric data along with maximum surface settlement are selected to evaluate the prediction performance. Comparative analysis shows that CNN-LSTM outperforms LSTM alone, while the TimeGAN-CNN-LSTM model provides superior accuracy over CNN-LSTM. In addition, experiments adjusting the training and test data ratios reveal that the TimeGAN-CNN-LSTM model significantly improves prediction accuracy, particularly when the training-to-test set ratio is 4-8.

Key words： shield tunnel; surface settlement; convolutional neural networks; long short-term memory networks; time series generative adversrial networks

Options

文章导航

〈

〉

模态框（Modal）标题

摘要

本文引用格式

Abstract