A deep-learning method for evaluating shaft resistance of the cast-in-site pile on reclaimed ground using field data  

基于现场试验的复垦地层灌注桩侧摩阻力的深度学习评价方法

作　　者：Sheng-liang LU Ning ZHANG Shui-long SHEN Annan ZHOU Hu-zhong LI 

机构地区：[1]Department of Civil and Environmental Engineering,College of Engineering,Shantou University,Shantou 515063,China [2]Key Laboratory of Intelligence Manufacturing Technology of Ministry of Education,Shantou University,Shantou 515063,China [3]Department of Architecture and Civil Engineering,Wenzhou Vocational and Technical College,Wenzhou 325000,China [4]Discipline of Civil and Infrastructure,School of Engineering,Royal Melbourne Institute of Technology(RMIT),Victoria 3001,Australia [5]Wenzhou Huabang Engineering Project Management Co.,Ltd.,Wenzhou 325207,China

出　　处：《Journal of Zhejiang University-Science A(Applied Physics & Engineering)》2020年第6期496-508,共13页浙江大学学报（英文版）A辑（应用物理与工程）

基　　金：the Research Funding of Shantou University for New Faculty Member(No.NTF19024-2019);the National Nature Science Foundation of China(No.41372283)。

摘　　要：This study proposes a deep learning-based approach for shaft resistance evaluation of cast-in-site piles on reclaimed ground,independent of theoretical hypotheses and engineering experience.A series of field tests was first performed to investigate the characteristics of the shaft resistance of cast-in-site piles on reclaimed ground.Then,an intelligent approach based on the long short term memory deep-learning technique was proposed to calculate the shaft resistance of the cast-in-site pile.The proposed method allows accurate estimation of the shaft resistance of cast-in-site piles,not only under the ultimate load but also under the working load.Comparisons with empirical methods confirmed the effectiveness of the proposed method for the shaft resistance estimation of cast-in-site piles on reclaimed ground in offshore areas.目的:基于极限平衡理论和诸多简化原则的经验公式方法难以适用于复杂的复垦地层中灌注桩的侧摩阻力计算。本文旨在探讨复垦地层中灌注桩在静力加载条件下的侧摩阻力发展规律和特性,并应用深度学习方法,以提高灌注桩侧摩阻力的预测精度。创新点:1.设计现场试验,研究近海复垦地层中灌注桩的承载能力特性;2.建立深度学习预测模型,高精度预测工作荷载下灌注桩的轴力和侧摩阻力。方法:1.通过实验分析,探明复垦地层中不同土层与桩体的相互作用和桩体侧摩阻力的发展规律;2.通过理论计算,指出经验方法在复垦地层灌注桩承载力计算中的缺陷和不足;3.通过序列化的人工智能方法建模,利用土体物理力学参数和桩身试验实测数据,对比验证深度学习方法的精度和计算效率。结论:1.灌注桩适用于复垦地层,能够为基础设施提供足够的承载力;2.经验方法对灌注桩中部桩体的极限侧摩阻力估计良好,而对地层条件较差的桩身两端的估计则存在较大偏差;3.深度学习方法能够综合考虑地层和桩体的相互作用,并且能精确预测在不同工作荷载和极限荷载下的侧摩阻力和桩身轴力,因而适用性更广。

关 键 词：Deep-learning method Cast-in-site pile Shaft resistance Field test Reclaimed ground 

分 类 号：TU473.1[建筑科学—结构工程]