All Issue

2024 Vol.40, Issue 6 Preview Page
Comparison of Deep Learning and Measurement-Based Methods for Settlement Prediction in Soft Ground

연약지반 침하예측을 위한 딥러닝 및 계측기반 기법의 예측 정확도 비교

Seongho Hong1
,
Tae-Young Kwak2
,
Sang-Inn Woo3
,
Sung-Ryul Kim4*
홍 성호1
,
곽 태영2
,
우 상인3
,
김 성렬4*
1Member, Ph.D Student, Dept. of Civil & Environmental Engrg., Seoul National Univ.
2Member, Senior Researcher, Dept. of Geotechnical Engrg. Research, Korea Institute of Civil Engineering and Building Technology
3Member, Assoc. Prof., Dept. of Civil and Environmental Engrg., Incheon National Univ.
4Member, Prof., Dept. of Civil & Environmental Engrg., Seoul National Univ.
1정회원, 서울대학교 건설환경공학부 박사과정
2정회원, 한국건설기술연구원 지반연구본부 수석연구원
3정회원, 인천대학교 도시환경공학부 부교수
4정회원, 서울대학교 건설환경공학부 교수
*Corresponding Author
31 December 2024. pp. 67-78
PDF XML Citation
Abstract

The accurate prediction of consolidation settlement is essential for construction sites on soft clay to optimize preload removal and reduce residual settlement. In Korea, measurement-based methods are commonly employed; however, these methods can be inaccurate with limited data and vary depending on the analysis period. Weighted nonlinear regression analysis and various deep learning models have been proposed to overcome these limitations; however, comparative analysis of these techniques remains insufficient. Thus, this study compares four deep learning algorithms (i.e., the ANN, LSTM, GRU, and Transformer models) and three measurement-based settlement prediction techniques (i.e., the hyperbolic method, Asaoka method, and weighted nonlinear regression analysis) across 392 conditions based on training and regression days (60-150 days). The results demonstrated that the weighted nonlinear regression analysis and GRU model generally obtained the highest prediction accuracy under all conditions, and the prediction accuracy of all techniques improved as the duration of the measurement data increased. When using 150 days of data, all the techniques achieved errors of less than 3 cm, providing accurate prediction results.

Keywords
Consolidation settlement
Deep learning
Neural networks
Settlement prediction
Time-series forecasting

대심도 연약지반에 선행재하 공법을 적용하는 경우 재하토 제거 시점을 예측하고 잔류침하량을 최소화하기 위해 연약지반의 침하거동을 정밀히 예측하는 것이 중요하다. 국내에서는 일반적으로 계측기반 침하예측 기법을 적용하고 있으나, 장기간 계측 결과가 필요하고 분석구간에 따라 예측이 달라지는 한계가 있다. 기존 침하예측 기법들의 한계를 보완하기 위해 가중 비선형 회귀 쌍곡선법과 여러 딥러닝 기반 최신 기법 및 모델들이 제시되었으나, 기법들 간의 비교·분석이 부족한 실정이다. 그러므로, 본 연구에서는 최근 제안된 딥러닝 모델들과 계측기반 침하예측 기법들의 정확도를 비교·분석하기 위해, 4개의 딥러닝 알고리즘(ANN, LSTM, GRU, Transformer)과 3개의 계측기반 침하예측 기법(쌍곡선법, Asaoka법, 가중 비선형 회귀 쌍곡선법)을 적용하여 학습 및 회귀 일수(60일-150일)에 따라 총 392개 조건에서 침하예측을 수행하였다. 분석 결과, 가중 비선형 회귀 쌍곡선법과 GRU 모델은 모든 조건에서 전반적으로 가장 높은 예측 정확도를 나타내었고 계측 데이터 사용 기간이 증가할수록 모든 기법의 예측 정확도가 향상되었다. 150일간의 데이터를 사용할 경우 모든 기법에서 3cm 이하의 오차를 달성하여 정확한 예측 결과를 제공하였다.

References
1

Asaoka, A. (1978), "Observational Procedure of Settlement Prediction", Soils and Foundations, Vol.18, No.4, pp.87-101.

10.3208/sandf1972.18.4_87
2

Barron, R. A. (1948), "Consolidation of Fine-grained Soils by Drain Wells by Drain Wells", Transactions of the American Society of Civil Engineers, Vol.113, No.1, pp.718-742.

10.1061/TACEAT.0006098
3

Chen, X. X., Yang, J., He, G. F., and Huang, L. C. (2023a), "Development of an LSTM-based Model for Predicting the Long-term Settlement of Land Reclamation and a GUI-based Tool", Acta Geotechnica, Vol.18, No.7, pp.3849-3862.

10.1007/s11440-022-01749-5
4

Chen, X., Yang, L., Xue, H., Li, L., and Yu, Y. (2023b), "A Machine Learning Model Based on GRU and LSTM to Predict the Environmental Parameters in a Layer House, Taking CO2 Concentration as an Example", Sensors, Vol.24, No.1, 244.

10.3390/s2401024438203104PMC10781361
5

Chen, Z. J., Feng, W. Q., Yin, J. H., and Shi, X. S. (2023c), "Finite Element Model and Simple Method for Predicting Consolidation Displacement of Soft Soils Exhibiting Creep Underneath Embankments in 2-D Condition", Acta Geotechnica, Vol.18, No.5, pp.2513-2528.

10.1007/s11440-022-01741-z
6

Cho, K. (2014), "Learning Phrase Representations Using RNN Encoder-decoder for Statistical Machine Translation", arXiv preprint arXiv:1406.1078.

10.3115/v1/D14-1179
7

Díaz, E. and Spagnoli, G. (2024), "A Super-learner Machine Learning Model for a Global Prediction of Compression Index in Clays", Applied Clay Science, Vol.249, 107239.

10.1016/j.clay.2023.107239
8

Ge, Q., Xia, Y., Shu, J., Li, J., and Sun, H. (2024), "Explainable Ensemble Learning Approaches for Predicting the Compression Index of Clays", Journal of Marine Science and Engineering, Vol.12, No.10, 1701.

10.3390/jmse12101701
9

Hansbo, S. (1960), "Consolidation of Clay with Special Reference to Influence of Vertical Sand Drains", Swedish Geotechnical Institute Proceeding, Vol.18, pp.45-50.

10

Hochreiter, S. (1997), "Long Short-term Memory", Neural Computation MIT-Press.

10.1162/neco.1997.9.8.17359377276
11

Hong, S., Ko, S. J., Woo, S. I., Kwak, T. Y., and Kim, S. R. (2024), "Time-series Forecasting of Consolidation Settlement Using LSTM Network", Applied Intelligence, Vol.54, No.2, pp.1386-1404.

10.1007/s10489-023-05219-7
12

Hoshino, S. (1962), "Problems of Foundation s in Recent Years", Society of Civil Engineering, Vol.47, No.7, pp.63-67 (in Japanese).

13

Kanayama, M., Rohe, A., and van Paassen, L. A. (2014), "Using and Improving Neural Network Models for Ground Settlement Prediction", Geotechnical and Geological Engineering, Vol.32, pp.687-697.

10.1007/s10706-014-9745-8
14

Kwak, T. Y., Hong, S., Lee, J. H., and Woo, S. I. (2022), "Analysis of the Limitations of the Existing Subsidence Prediction Method Based on the Subsidence Measurement Data and Suggestions for Improvement Method Through Weighted Nonlinear Regression Analysis", Journal of the Korean Geotechnical Society, Vol.38, No.12, pp.103-112 (in Korean).

15

Li, Z., Peng, Y., Li, J., and Tang, Z. (2024), "Composite Foundation Settlement Prediction Based on LSTM-Transformer Model for CFG", Applied Sciences, Vol.14, No.2, 732.

10.3390/app14020732
16

Lo, M. K., Loh, D. R., Chian, S. C., and Ku, T. (2023), "Probabilistic Prediction of Consolidation Settlement and Pore Water Pressure Using Variational Autoencoder Neural Network", Journal of Geotechnical and Geoenvironmental Engineering, Vol.149, No.1, 04022119.

10.1061/JGGEFK.GTENG-10555
17

Mikasa, M. (1963), "Consolidation of Soft Clay", Kajima-shuppan-kai, Tokyo, Japan (in Japanese).

18

Miller, J. A., Aldosari, M., Saeed, F., Barna, N. H., Rana, S., Arpinar, I. B., and Liu, N. (2024), "A Survey of Deep Learning and Foundation Models for Time Series Forecasting", arXiv preprint arXiv:2401.13912.

19

Monden, H. (1963), A new time-fitting method for the Settlement analySiS Of fOundation On SOft ClayS, Memoir FaClty Of Engrg, Hiroshima Univ., 2-1.

20

Rumelhart, D. E., Hinton, G. E., and Williams, R. J. (1986), "Learning Representations by Back-propagating Errors", Nature, Vol.323, No.6088, pp.533-536.

10.1038/323533a0
21

Sridharan, A., Murthy, N. S., and Prakash, K. (1987), "Rectangular Hyperbola Method of Consolidation Analysis", Geotechnique, Vol.37, No.3, pp.355-368.

10.1680/geot.1987.37.3.355
22

Tan, S. A. (1993), "Ultimate Settlement by Hyperbolic Plot for Clays with Vertical Drains", Journal of Geotechnical Engineering, ASCE, Vol.119, No.5, pp.950-956

10.1061/(ASCE)0733-9410(1993)119:5(950)
23

Tan, S. and Chew, S. (1996), "Comparison of the Hyperbolic and Asaoka Observational Method of Monitoring Consolidation with Vertical Drains", Soils Found, Vol.36, pp.31-42.

10.3208/sandf.36.3_31
24

Tan, T. S., Inoue, T., and Lee, S. L. (1991), "Hyperbolic Method for Consolidation Analysis", Journal of Geotechnical Engineering, Vol.117, No.11, pp.1723-1737.

10.1061/(ASCE)0733-9410(1991)117:11(1723)
25

Terzaghi, K. (1943), Theoretical soil mechanics.

10.1002/9780470172766
26

Vaswani, A. (2017), "Attention is All You Need", Advances in Neural Information Processing Systems.

27

Yoo, H. K. and Kim, J. H. (2000), "A Study on the Prediction of Long-Term Settlement by the Modified Hyperbolic Method", Journal of the Korean Geotechnical Society, Vol.16, No.3, pp.163-172 (in Korean).

28

Zhang, G. P. and Kline, D. M. (2007), "Quarterly Time-series Forecasting with Neural Networks", IEEE Transactions on Neural Networks, Vol.18, No.6, pp.1800-1814.

10.1109/TNN.2007.896859
Information
  • Publisher :The Korean Geotechnical Society
  • Publisher(Ko) :한국지반공학회
  • Journal Title :Journal of the Korean Geotechnical Society
  • Journal Title(Ko) :한국지반공학회 논문집
  • Volume : 40
  • No :6
  • Pages :67-78
  • Received Date : 2024-10-30
  • Accepted Date : 2024-11-17
  • DOI :https://doi.org/10.7843/kgs.2024.40.6.67
Journal Informaiton Journal of the Korean Geotechnical Society Journal of the Korean Geotechnical Society
Archives
: Vol.23~28 (2007~2012)
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close