在基础坑工程中用于管道变形预测的时空相关性和基于注意力的模型
Wanghu Chen1, Shi Yuan2, Lei He2
1College of Computer Science and Engineering, Northwest Normal University, Lanzhou, 730070, China. chenwh@nwnu.edu.cn.
Scientific reports
|November 3, 2024
概括
预测基础坑中的相邻管道变形对于安全至关重要. 使用卷积神经网络和双向长短内存单元的新深度学习模型显著提高了预测准确性和长期预测.
科学领域:
- 地质技术工程 地质技术工程
- 人工智能的人工智能
- 土木工程 土木工程是指土木工程.
背景情况:
- 在基础坑工程中,对相邻管道的变形预测对于建筑安全至关重要.
- 由于复杂的条件和非静止,非线性监测数据,现有的方法 (构成模型,灰色相关性,前神经网络) 面临挑战.
- 准确的预测对于防止结构损坏和确保项目的完整性至关重要.
研究的目的:
- 为基础坑中的相邻管道变形提出一种基于深度学习的新型预测模型.
- 解决现有模型在处理复杂的水文,地质和数据特征方面的局限性.
- 提高变形预测的准确性和可靠性,特别是在长期情景中.
主要方法:
- 拟定管道变形作为一个多变量时间序列问题.
- 开发了一个深度学习模型,将卷积神经网络 (CNN) 集成为空间依赖提取和双向长期短期记忆 (BiLSTM) 单元用于时间特征提取.
- 整合了注意力机制,以动态调整时空特征的权重.
主要成果:
- 拟议的模型在现实世界地铁项目评估中,与现有方法相比,表现优越.
- 在预测准确度方面取得了显著的改进,调整后R2指数从19.4增加到61.6.
- 与其他模型相比,平均绝对误差显著减少,减少了51.5至70.3.
- 验证了捕捉时空依赖和注意力学习的有效性,以改进工程预测.
结论:
- 深度学习模型有效地捕捉了监控数据中的复杂的时空依赖关系.
- 注意力机制增强了模型优先考虑相关特征的能力,以准确预测变形.
- 拟议的方法为预测基础坑工程中的相邻管道变形提供了重大进步,特别是在长期预测方面.
更多相关视频
06:55Kinematic History of a Salient-recess Junction Explored through a Combined Approach of Field Data and Analog Sandbox Modeling
Published on: August 5, 2016
8.2K
07:58Data Processing Methods for 3D Seismic Imaging of Subsurface Volcanoes: Applications to the Tarim Flood Basalt
Published on: August 7, 2017
9.3K
相关概念视频
Field Procedure for Staking Out Curves
33
Staking out curves is an essential process in construction to ensure the accurate alignment of structures along a curved path. This task involves positioning stakes at calculated locations corresponding to the curve's design, effectively translating plans into physical markers in the field. The process begins by determining the geometric parameters of the curve, including the radius, central angle, and tangent distances. These parameters are critical for identifying key points such as the...
33
Design Example: Creating a Hydraulic Model of a Dam Spillway
127
Scaled hydraulic models of dam spillways provide a practical way to replicate and study the intricate flow dynamics of these structures. Often built to a 1:15 ratio, these models allow for observing critical water behavior, such as velocity distribution, flow patterns, and energy dissipation.
127
Typical Model Studies
340
Fluid mechanics model studies often utilize scaled-down systems to predict fluid behavior in full-scale environments, such as river flows, dam spillways, and structures interacting with open surfaces. Maintaining Froude number similarity in river models is crucial, as it replicates surface flow features like wave patterns and velocities.
340
