基于神经网络的板墙组合式固化土地基承载力计算方法与优化设计

周灵刚; 胡奕挺; 陈欣蔚; 屠锋; 吴朝峰; 于洋; 王彦兵; 满银; 李维朝

doi:10.19509/j.cnki.dzkq.tb20230720

基于神经网络的板墙组合式固化土地基承载力计算方法与优化设计

doi: 10.19509/j.cnki.dzkq.tb20230720

周灵刚^1,,
胡奕挺¹,
陈欣蔚²,
屠锋³,
吴朝峰⁴,
于洋^2, ,,
王彦兵⁵,
满银⁶,
李维朝⁷

1.
国网浙江省电力有限公司台州供电公司, 浙江台州 318001
2.
浙江大学海洋学院, 浙江舟山 316021
3.
国网浙江省电力有限公司, 杭州 310007
4.
中国能源建设集团浙江省电力设计院有限公司, 杭州 310014
5.
国网经济技术研究院有限公司, 北京 102200
6.
中国电力科学研究院有限公司, 北京 100192
7.
中国水利水电科学研究院, 北京 100038

基金项目:

国家电网依托工程科技项目 11CJ01

详细信息

作者简介:
周灵刚, E-mail: 31558565@qq.com

通讯作者:
于洋, E-mail: yang-yu@zju.edu.cn

中图分类号: TU470
计量
- 文章访问数: 92
- PDF下载量: 16
- 被引次数: 0
出版历程
- 收稿日期: 2023-12-28
- 录用日期: 2024-08-05
- 修回日期: 2024-08-03

Calculation of capacity and optimization design-composite slab wall soil solidification foundation based on neural network

1.
State Grid Zhejiang Taizhou Power Supply Company, Taizhou Zhejiang 318001, China
2.
Ocean College, Zhejiang University, Zhoushan Zhejiang 316021, China
3.
State Grid Zhejiang Electric Power Company, Hangzhou 310007, China
4.
China Engineering Group Zhejiang Power Design Institute Co., Ltd., Hangzhou 310014, China
5.
State Power Economic Research Institute Co. Ltd., Beijing 102200, China
6.
China Electric Power Research Institute, Beijing 100192, China
7.
China Institute of Water Resources and Hydropower Research, Beijing 100038, China

More Information

Author Bio:
ZHOU Linggang, E-mail: 31558565@qq.com

Corresponding author: YU Yang, E-mail: yang-yu@zju.edu.cn

摘要

摘要:
软土固化技术在地基处理中应用广泛, 为发挥其空间可塑性优势, 板墙组合式固化土地基逐步在工程中得到应用。然而, 板墙组合式固化土地基承载力尚无可靠计算方法, 叠加固化土与软土力学参数不确定性, 导致板墙组合式固化土地基优化设计困难, 需要提出一种板墙组合式固化土地基承载力计算及优化设计方法。以浙江省台州市经纬110 kV滨海变电站场地为研究对象, 测试固化土与软土力学参数, 建立数值模型计算板墙组合式地基承载力, 以此为基础训练神经网络模型, 作为板墙组合式固化土地基承载力计算模型, 为工程应用提供便利。引入岩土工程鲁棒性设计理论, 采用蒙特卡洛模拟处理固化土与软土力学参数不确定性, 以标准差作为鲁棒性评价指标评估不确定性对设计的影响, 以固化土地基截面面积近似表征工程造价, 实现考虑经济性和鲁棒性的固化土地基优化设计。将设计方法应用于工程实例, 得到最优设计方案为固化板厚度P=2 m、固化墙深度W=3 m、固化墙厚度D=1.5 m、固化墙净间距S=1 m, 为工程设计提供参考。本研究提出的板墙组合式固化土地基承载力计算模型及优化设计方法为相关工程计算和设计提供新思路和新方法。
- 固化土地基 /
- 神经网络 /
- 承载力 /
- 鲁棒性设计 /
- 软土固化 /
- 板墙组合
Abstract: Objective
The soil solidification technique is widely used in soft foundation treatment. To exploit spatial plasticity of this technique, composite slab wall soil solidification foundations have gradually been applied in engineering projects. However, a reliable method for calculating the bearing capacity of composite slab wall soil solidification foundations is lack, and the mechanical parameters of both solidified and soft soil remain uncertain. These factors complicate the optimization of the composite slab wall soil solidification foundation designs. Therefore, it is crucial to propose a method for calculating the bearing capacity and optimizing the design of such foundations.
Methods
This study focuses on the 110 kV Jingwei coastal substation in Taizhou, Zhejiang Province. A numerical model is established based on the mechanical parameters of solidified and soft soil to calculate the bearing capacity of composite slab wall soil solidification foundations. The results of these calculations are used to train a neural network, enabling predictions of the bearing capacity for various design parameters, thus facilitating engineering applications. Uncertainties of the mechanical parameters are addressed through Monte Carlo simulations, and their impact on design is estimated using the robustness evaluation index standard deviation. The design cost is approximately estimated by the cross-sectional area of the foundation. Robust design theory is introduced to optimize the design while balancing cost-effectiveness and robustness.
Results
This method is implemented in an engineering project, resulting in an optimal design with solidified plate thickness P=2 m, solidified wall depth W=3 m, solidified wall thickness D=1.5 m, solidified wall net spacing S=1 m, and upper foundation width B=4 m, providing a reference for engineering designs.
Conclusion
The proposed methods for calculating bearing capacity and optimizing the design of composite slab wall soil solidification foundations offer new concepts and approaches for similar projects.
- soil solidification foundation /
- neural network /
- bearing capacity /
- robust design /
- soft soil solidification /
- composite slab wall
注释:

所有作者声明不存在利益冲突。

HTML全文

图 1 工程地理位置

Figure 1. Position of the study area

下载: 全尺寸图片幻灯片

图 2 场地地层分布图

Figure 2. Stratigraphic distribution of the study site

下载: 全尺寸图片幻灯片

图 3 板墙组合式固化土地基示意图

W.固化墙深度；P.固化板厚度；B.上部基础宽度；D.固化墙厚度；S.固化墙净间距；下同

Figure 3. Diagram of composite slab wall soil solidification foundation

下载: 全尺寸图片幻灯片

图 4 板墙组合式固化土地基数值模型与计算结果

a.数值模型; b.基础底面压力变化曲线；○表示只限制了法向位移；△表示限制了xy 2个方向的位移

Figure 4. Numerical model and calculation results of composite slab wall soil solidification foundation

下载: 全尺寸图片幻灯片

图 5 神经网络结构示意图

Figure 5. Schematic of the neural network structure

下载: 全尺寸图片幻灯片

图 6 帕累托前沿和关节点示意图

Figure 6. Schematic of Pareto front and knee point

下载: 全尺寸图片幻灯片

图 7 板墙组合式固化土地基设计流程图

Figure 7. Design flowchart of composite slab wall soil solidification foundation

下载: 全尺寸图片幻灯片

图 8 神经网络拟合结果

Figure 8. Fitting results of neural network

下载: 全尺寸图片幻灯片

图 9 不同噪声因素组数对30个噪声因素集合的变异系数(COV)的影响

Figure 9. Influence of the number of noise factors on the coefficient of variation for 30 collections of noise factors

下载: 全尺寸图片幻灯片

图 10 地基承载力分布直方图

Figure 10. Distribution histogram of bearing capability of foundations

下载: 全尺寸图片幻灯片

图 11 板墙组合式固化土地基鲁棒性设计的帕累托前沿及关节点

Figure 11. Pareto front and knee point of the robust design for the composite slab wall soil solidification foundation

下载: 全尺寸图片幻灯片

图 12 土体参数不确定性变化后的帕累托前沿和关节点

Figure 12. Pareto front and knee points for variation in the uncertainties of soil parameters

下载: 全尺寸图片幻灯片

图 13 Case 1、Case 2和Case 3的帕累托前沿和关节点

Figure 13. Pareto front and knee point for Case 1, Case 2 and Case 3

下载: 全尺寸图片幻灯片

表 1 土体物理力学参数

Table 1. Physical and mechanical parameters of soil

地层	重度γ/ (kN·m^-3)	压缩摸量E_s/MPa	黏聚力c/kPa	内摩擦角φ(°)	渗透系数k/(cm·s^-1)
冲填土(吹填土)	17.3	2.6	12.3	10.2	1.18 × 10^-5
淤泥质粉质黏土	17.7	3.0	13.7	12.1	8.47 × 10^-6
淤泥质黏土	17.0	2.3	12.1	9.5	5.26 × 10^-6
粉质黏土	17.9	3.4	22.3	13.3	9.62 × 10^-6
黏土	17.7	3.4	24.4	13.3	6.47 × 10^-6
粉质黏土	19.2	5.8	36.5	18.0	7.33 × 10^-6
固化土	18.0	4.26	50.0	31.6	1.54 × 10^-4

下载: 导出CSV

表 2 神经网络训练样本输入参数的取值

Table 2. Input parameters of the neural network training samples

参数	最小值	平均值	最大值
基础宽度B/m	2	4	6
固化墙深度W/m	4	6	8
固化板厚度P/m	2	4	6
固化墙厚度D/m	0.5	1.0	1.5
固化墙净间距S/m	1	2	3
固化土黏聚力c₁/kPa	30	40	50
固化土内摩擦角φ₁/(°)	21	28	35
软土黏聚力c₂/kPa	10	15	20
软土内摩擦角φ₂/(°)	9	12	15

下载: 导出CSV

表 3 设计参数取值

Table 3. Design parameters values

设计参数	取值
基础宽度B/m	4
固化墙深度W/m	3，4，5，6，7，8
固化板厚度P/m	2，3，4，5，6
固化墙厚度D/m	1.0，1.5
固化墙净间距S/m	1，2，3

下载: 导出CSV

表 4 不确定性参数取值情况

Table 4. Noise factors values

土体种类	参数	统计量	取值	分布形式
下伏软土	黏聚力c₂	均值/kPa	13.7	对数正态分布
	黏聚力c₂	变异系数COV	0.3
	内摩擦角φ₂	均值/(°)	12.1
	内摩擦角φ₂	变异系数COV	0.3
固化土	黏聚力c₁	均值/kPa	50.0
	黏聚力c₁	变异系数COV	0.1
	内摩擦角φ₁	均值/(°)	31.6
	内摩擦角φ₁	变异系数COV	0.1

下载: 导出CSV

表 5 帕累托前沿上各设计点鲁棒性和工程造价归一化结果及到乌托邦点的距离

Table 5. Distance between the normalized robustness, the normalized cost of the design and knee point on the Pareto front

序号	设计参数							归一化		距离
序号	B/m	P/m	W/m	S/m	D/m	SNR	C/m²	SNR	C	距离
1	4	2	3	3	1	3.260	31	0.000	0.000	1.000
2	4	2	3	2	1	3.342	32	0.149	0.017	0.852
3	4	2	3	2	1.5	3.422	34	0.294	0.050	0.708
4	4	2	3	1	1	3.437	35	0.322	0.066	0.682
5	4	2	3	1	1.5	3.521	35.5	0.475	0.074	0.531
6	4	3	4	1	1.5	3.541	49.5	0.511	0.306	0.577
7	4	4	5	1	1.5	3.631	63.5	0.675	0.537	0.628
8	4	5	6	1	1	3.636	77	0.682	0.760	0.824
9	4	5	6	1	1.5	3.700	77.5	0.799	0.769	0.794
10	4	6	7	1	1	3.711	91	0.818	0.992	1.008
11	4	6	7	1	1.5	3.811	91.5	1.000	1.000	1.000
注: C.工程造价(固化土面积); SNR.鲁棒性(信噪比)

下载: 导出CSV

表 6 参数不确定性变化情况

Table 6. Variation in the uncertainties of soil parameters

组别	下伏软土				固化土				分布形式
	黏聚力c₂		内摩擦角φ₂		黏聚力c₁		内摩擦角φ₁
	均值/kPa	COV	均值/(°)	COV	均值/kPa	COV	均值/(°)	COV
Case 1		0.2		0.2		0. 05		0. 05	对数正态分布
Case 2	13.7	0.3	12.1	0.3	50. 0	0. 10	31. 6	0. 10
Case 3		0.4		0.4		0. 15		0. 15

下载: 导出CSV

表 7 Case 1、Case 2和Case 3的最优设计的设计参数、鲁棒性和工程造价

Table 7. Design parameters, design robustness and cost of optimal design for Case 1, Case 2 and Case3

组别	设计参数					鲁棒性(信噪比SNR)	工程造价(固化土面积C)/m²
组别	B/m	P/m	W/m	S/m	D/m	鲁棒性(信噪比SNR)	工程造价(固化土面积C)/m²
Case 1	4	2	3	1	1.5	4.124	35.5
Case 2	4	2	3	1	1.5	3.521	35.5
Case 3	4	2	3	1	1.5	3.268	35.5

下载: 导出CSV

参考文献(21)

[1]	蔡强. 滨海深厚软土加固变形机理及工程运用[D]. 安徽淮南: 安徽理工大学, 2018. CAI Q. Deformation mechanism and engineering application of the strengthened deep soft soil in the coastal[D]. Huainan Anhui: Anhui University of Science and Technology, 2018. (in Chinese with English abstract)
[2]	王冲, 郝金山, 张迎, 等. 天津滨海地区软土固化地基承载性状分析[J]. 勘察科学技术, 2015(2): 1-4. WANG C, HAO J S, ZHANG Y, et al. Analysis on bearing behavior of soft soil solidified foundation in Tianjin coastal area[J]. Site Investigation Science and Technology, 2015(2): 1-4. (in Chinese with English abstract)
[3]	张吉, 赵丹禄. 天津滨海地区软土固化技术现场试验研究[J]. 珠江水运, 2023(9): 104-106. ZHANG J, ZHAO D L. Field test study on soft soil solidification technology in coastal area of Tianjin[J]. Pearl River Water Transport, 2023(9): 104-106. (in Chinese with English abstract)
[4]	刘曦. 带肋梁硬壳层软土地基应力分布与承载力研究[D]. 沈阳: 沈阳建筑大学, 2013. LIU X. Research on the stress distribution and bearing capacity of the hard shell foundation with rib beam[D]. Shenyang: Shenyang Architecture University, 2013. (in Chinese with English abstract)
[5]	王玉梅. 带肋梁硬壳层软土地基应力与变形的研究[D]. 沈阳: 沈阳建筑大学, 2013. WANG Y M. Research on the stress and deformation of the hard shell foundation with rib beam[D]. Shenyang: Shenyang Architecture University, 2013. (in Chinese with English abstract)
[6]	LIU X Q, ZHANG H Q, YAN Z, et al. Research on bearing characteristics of grid composite foundation based on silt solidification[J]. Journal of Performance of Constructed Facilities, 2021, 35(5): 04021070. doi: 10.1061/(ASCE)CF.1943-5509.0001611
[7]	曹文庚, 潘登, 徐郅杰, 等. 河南省滑坡灾害易发性制图研究: 多种机器学习模型的对比[J/OL]. 地质科技通报: 1-11(2023-12-28)[2024-09-02]. doi: 10.19509/j.cnki.dzkq.tb20230338. CAO W G, PAN D, XU Z J, et al. Landslide disaster vulnerability mapping study in Henan Province: Comparison of different machine learning models[J/OL]. Bulletin of Geological Science and Technology: 1-11(2023-12-28)[2024-09-02]. doi: 10.19509/j.cnki.dzkq.tb20230338.(in Chinese with English abstract)
[8]	宋宜祥, 张晓波, 黄达. 基于SSA-Adam-BP神经网络模型的堰塞坝稳定性预测[J]. 地质科技通报, 2022, 41(2): 130-138. doi: 10.19509/j.cnki.dzkq.2022.0040 SONG Y X, ZHANG X B, HUANG D. Stability prediction of landslide dams based on SSA-Adam-BP neural network model[J]. Bulletin of Geological Science and Technology, 2022, 41(2): 130-138. (in Chinese with English abstract) doi: 10.19509/j.cnki.dzkq.2022.0040
[9]	邓日朗, 张庆华, 刘伟, 等. 基于改进两步法采样策略和卷积神经网络的崩塌易发性评价[J]. 地质科技通报, 2024, 43(2): 186-200. doi: 10.19509/j.cnki.dzkq.tb20220535 DENG R L, ZHANG Q H, LIU W, et al. Collapse susceptibility evaluation based on an improved two-step sampling strategy and a convolutional neural network[J]. Bulletin of Geological Science and Technology, 2024, 43(2): 186-200. (in Chinese with English abstract) doi: 10.19509/j.cnki.dzkq.tb20220535
[10]	郭衍昊, 窦杰, 向子林, 等. 基于优化负样本采样策略的梯度提升决策树与随机森林的汶川同震滑坡易发性评价[J]. 地质科技通报, 2024, 43(3): 251-265. doi: 10.19509/j.cnki.dzkq.tb20230037 GUO Y H, DOU J, XIANG Z L, et al. Susceptibility evaluation of Wenchuan coseismic landslides by gradient boosting decision tree and random forest based on optimal negative sample sampling strategies[J]. Bulletin of Geological Science and Technology, 2024, 43(3): 251-265. (in Chinese with English abstract) doi: 10.19509/j.cnki.dzkq.tb20230037
[11]	宋居正, 刘思位, 吴云山. 基于前馈神经网络的外挂物投放分离预测[J]. 弹道学报, 2023, 35(3): 33-38. SONG J Z, LIU S W, WU Y S. External store separation prediction based on feedforward neural network[J]. Journal of Ballistics, 2023, 35(3): 33-38. (in Chinese with English abstract)
[12]	路晓宇. 大连海相软土固化强度试验研究及其神经网络预测[D]. 辽宁大连: 大连理工大学, 2021. LU X Y. Experimental study on solidification strength of Dalian marine soft soil and its neural network prediction[D]. Dalian Liaoning: Dalian University of Technology, 2021. (in Chinese with English abstract)
[13]	王才进, 武猛, 杨洋, 等. 基于生物地理优化的人工神经网络模型预测软土的固结系数[J]. 岩土力学, 2023, 44(10): 3022-3030. WANG C J, WU M, YANG Y, et al. Prediction of consolidation coefficient of soft soil using an artificial neural network models with biogeography-based optimization[J]. Rock and Soil Mechanics, 2023, 44(10): 3022-3030. (in Chinese with English abstract)
[14]	丁建文, 魏霞, 高鹏举, 等. 基于GA-BP神经网络的软土路基运营期沉降预测[J]. 东南大学学报(自然科学版), 2023, 53(4): 585-591. DING J W, WEI X, GAO P J, et al. Prediction of settlement of soft soil subgrade during operation based on GA-BP neural network[J]. Journal of Southeast University (Natural Science Edition), 2023, 53(4): 585-591. (in Chinese with English abstract)
[15]	JUANG C H, LIU Z F, ATAMTURKTUR H S. Reliability-based robust geotechnical design of retaining walls[C]//Anon. Sound Geotechnical Research to Practice. San Diego, California, USA. Reston, VA: American Society of Civil Engineers, 2013: 514-524.
[16]	JUANG C H, WANG L, LIU Z F, et al. Robust geotechnical design of drilled shafts in sand: New design perspective[J]. Journal of Geotechnical and Geoenvironmental Engineering, 2013, 139(12): 2007-2019. doi: 10.1061/(ASCE)GT.1943-5606.0000956
[17]	JUANG C H, WANG L, HSIEH H S, et al. Robust geotechnical design of braced excavations in clays[J]. Structural Safety, 2014, 49: 37-44. doi: 10.1016/j.strusafe.2013.05.003
[18]	胡海晨. 基于BP神经网络优化超声波测距的多功能除雪车远程监控系统研究与开发[D]. 沈阳: 东北大学, 2022. HU H C. Research and development of remote monitoring system for multifunctional snow removal vehicle based on BP neural network to optimize ultrasonic ranging[D]. Shenyang: Northeastern University, 2022. (in Chinese with English abstract)
[19]	吴选蓉. 基于GA-BP神经网络的煤岩组合破坏预警方法研究[D]. 重庆: 重庆大学, 2022. WU X R. Research on early warning method of coal and rock combination failure based on GA-BP neural network[D]. Chongqing: Chongqing University, 2022. (in Chinese with English abstract)
[20]	DAI X Q, SHI H B, LI Y S, et al. Artificial neural network models for estimating regional reference evapotranspiration based on climate factors[J]. Hydrological Processes, 2009, 23(3): 442-450. doi: 10.1002/hyp.7153
[21]	KHOSHNEVISAN S, GONG W P, WANG L, et al. Robust design in geotechnical engineering: An update[J]. Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 2014, 8(4): 217-234.