All Issue

2026 Vol.42, Issue 4 Preview Page
Depthwise LSTM-Based Stratigraphic Record Consistency Evaluation and Uncertainty Classification for SPT Data Quality Control

SPT 데이터 품질관리를 위한 심도 방향 LSTM 기반 지층 기록 일관성 평가 및 불확실성 유형화

Jeong-Sik Park1
,
Chang-Ho Shin2
,
Taeseo Ku3*
박 정식1
,
신 창호2
,
구 태서3*
1Member, Graduate Student, Dept. of Civil & Environmental Engineering, Konkuk Univ.
2Member, Graduate Student, Dept. of Civil & Environmental Engineering, Konkuk Univ.
3Member, Associate Prof., Dept. of Civil & Environmental Engineering, Konkuk Univ.
1정회원, 건국대학교 사회환경공학부 석사과정
2정회원, 건국대학교 사회환경공학부 석사과정
3정회원, 건국대학교 사회환경공학부 부교수
*Corresponding Author
31 August 2026. pp. 21-33
PDF XML Citation
Abstract

This study proposes a method for evaluating the consistency between standard penetration test (SPT) data and stratigraphic records while classifying intervals that require further inspection based on predictive uncertainty. Publicly available geotechnical investigation data from Seoul were analyzed, with four stratigraphic groups—Fill, Sediment, Weathered Soil, and Rock—serving as proxy labels. Stratigraphic classification served as a surrogate task to learn the underlying relationship between SPT input data and recorded stratigraphy. A unidirectional long short-term memory (LSTM) model was developed using depth, SPT N-value, penetration depth, and depthwise change indicators as input features, and the effectiveness of depthwise sequence information was examined through comparison with a fully connected neural network (FCNN). On the test set, the unidirectional LSTM achieved an accuracy of 72.76% and a macro F1 score of 0.6701, outperforming the FCNN. The entropy-based area under the receiver operating characteristic curve was 0.7340, indicating that predictive entropy was associated with misclassification detection. By combining prediction performance and normalized entropy, four consistency types were defined. High-entropy types were closely associated with layer boundaries and abrupt changes in N-values, whereas low-entropy misclassified types showed greater consistency with the empirical SPT distribution of the predicted stratigraphic group than with that of the recorded group. The proposed method provides a quality-control support tool for prioritizing intervals requiring expert review.

Keywords
Depthwise sequence
Long short-term memory
Quality control
Soil classification
Standard penetration test

본 연구에서는 대규모 표준관입시험 자료의 품질관리를 위해 표준관입시험 기록과 지층 기록 사이의 일관성을 평가하고, 예측 불확실성에 기반하여 검토 대상 구간을 유형화하는 방법을 제안하였다. 서울지역 지반조사 자료를 활용하여 매립층, 퇴적층, 풍화토층, 암반층의 4개 지층군을 참조 라벨로 정의하고, 표준관입시험 입력자료와 지층 기록 간의 지배적 관계를 학습하기 위한 대리 과제로 지층 분류 문제를 설정하였다. 심도, N값, 관입깊이 및 심도 방향 변화 지표를 입력변수로 하여 단방향 LSTM 모델을 구성하였으며, FCNN과의 비교를 통해 심도 방향 시퀀스 정보의 유효성을 검토하였다. 시험 자료에서 단방향 LSTM은 정확도 72.76%, macro F1 0.6701을 보여 FCNN보다 높은 성능을 나타냈으며, 예측 엔트로피 기반 AUC는 0.7340으로 산정되었다. 정분류 여부와 정규화 엔트로피를 결합하여 네 가지 유형을 정의한 결과, 고엔트로피 유형은 지층 경계부 및 N값 급변 구간과 관련성이 높았고, 저엔트로피 오분류 유형은 기록 지층군보다 예측 지층군의 경험적 SPT 분포와 더 높은 일관성을 보였다. 제안 방법은 전문가 검토 우선순위를 통한 품질관리 지원 도구로 활용될 수 있다.

References
1

Bengio, Y., Courville, A., and Vincent, P. (2013), “Representation Learning: A Review and New Perspectives”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.35, No.8, pp.1798-1828.

10.1109/TPAMI.2013.50
2

Christian, J. T. (2004), “Geotechnical Engineering Reliability: How Well Do We Know What We Are Doing?”, Journal of Geotechnical and Geoenvironmental Engineering, Vol.130, No.10, pp.985-1003, https://doi.org/10.1061/(ASCE)1090-0241(2004)130:10(985).

10.1061/(ASCE)1090-0241(2004)130:10(985)
3

Das, B. M. and Sivakugan, N. (2017), Fundamentals of Geotechnical Engineering, 5th Ed., Cengage Learning, Boston, pp.344-350.

4

Gawlikowski, J., Tassi, C. R. N., Ali, M., Lee, J., Humt, M., Feng, J., Kruspe, A., Triebel, R., Jung, P., Roscher, R., Shahzad, M., Yang, W., Bamler, R., and Zhu, X. X. (2023), “A Survey of Uncertainty in Deep Neural Networks”, Artificial Intelligence Review, Vol.56, Suppl.1, pp.1513-1589, https://doi.org/10.1007/s10462-023-10562-9.

10.1007/s10462-023-10562-9
5

Goodfellow, I., Bengio, Y., and Courville, A. (2016), Deep Learning, MIT Press, Cambridge, pp.363-408.

6

Hasancebi, N. and Ulusay, R. (2007), “Empirical Correlations between Shear Wave Velocity and Penetration Resistance for Ground Shaking Assessments”, Bulletin of Engineering Geology and the Environment, Vol.66, No.2, pp.203-213, https://doi.org/10.1007/s10064-006-0063-0.

10.1007/s10064-006-0063-0
7

Hatanaka, M. and Uchida, A. (1996), “Empirical Correlation between Penetration Resistance and Internal Friction Angle of Sandy Soils”, Soils and Foundations, Vol.36, No.4, pp.1-9, https://doi.org/10.3208/sandf.36.4_1.

10.3208/sandf.36.4_1
8

Heo, G. S. and Kwak, D. Y. (2022), “VS Prediction Model Using SPT-N Values and Soil Layers in South Korea”, Journal of the Korean Geotechnical Society, Vol.38, No.8, pp.53-66, https://doi.org/10.7843/kgs.2022.38.8.53.

10.7843/kgs.2022.38.8.53
9

Hodge, V. J. and Austin, J. (2004), “A Survey of Outlier Detection Methodologies”, Artificial Intelligence Review, Vol.22, No.2, pp.85-126, https://doi.org/10.1023/B:AIRE.0000045502.10941.a9.

10.1023/B:AIRE.0000045502.10941.a9
10

Hubert, M. and Vandervieren, E. (2008), “An Adjusted Boxplot for Skewed Distributions”, Computational Statistics & Data Analysis, Vol.52, No.12, pp.5186-5201, https://doi.org/10.1016/j.csda.2007.11.008.

10.1016/j.csda.2007.11.008
11

Hüllermeier, E. and Waegeman, W. (2021), “Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methods”, Machine Learning, Vol.110, No.3, pp.457-506, https://doi.org/10.1007/s10994-021-05946-3.

10.1007/s10994-021-05946-3
12

Kingma, D. P. and Ba, J. (2015), “Adam: A Method for Stochastic Optimization”, Proc. of the 3rd International Conference on Learning Representations (ICLR), San Diego, CA, USA, pp.1-15.

13

LeCun, Y., Bengio, Y., and Hinton, G. (2015), “Deep Learning”, Nature, Vol.521, No.7553, pp.436-444, https://doi.org/10.1038/nature14539.

10.1038/nature14539
14

Murthy, V. N. S. (2002), Geotechnical Engineering: Principles and Practices of Soil Mechanics and Foundation Engineering, CRC Press, Boca Raton, pp.327-330.

15

Ohta, Y. and Goto, N. (1978), “Empirical Shear Wave Velocity Equations in Terms of Characteristic Soil Indexes”, Earthquake Engineering & Structural Dynamics, Vol.6, No.2, pp.167-187, https://doi.org/10.1002/eqe.4290060205.

10.1002/eqe.4290060205
16

Park, K. H., Kim, J. K., Lee, S. H., Kim, M. K., Lee, K. R., and Han, J. T. (2022), “A Study on the Quality Control Method for Geotechnical Information Using AI”, Journal of the Korean Geotechnical Society, Vol.38, No.11, pp.87-95.

10.7843/KGS.2022.38.11.87
17

Phoon, K. K. and Kulhawy, F. H. (1999), “Characterization of Geotechnical Variability”, Canadian Geotechnical Journal, Vol.36, No.4, pp.612-624, https://doi.org/10.1139/t99-038.

10.1139/t99-038
18

Seoul Metropolitan Government (2006), Geotechnical Investigation Manual, Technical Review Officer, Seoul Metropolitan Government, Seoul, Korea, pp.6-17.

19

Skempton, A. W. (1986), “Standard Penetration Test Procedures and the Effects in Sands of Overburden Pressure, Relative Density, Particle Size, Ageing and Overconsolidation”, Géotechnique, Vol.36, No.3, pp.425-447, https://doi.org/10.1680/geot.1986.36.3.425.

10.1680/geot.1986.36.3.425
Information
  • Publisher :The Korean Geotechnical Society
  • Publisher(Ko) :한국지반공학회
  • Journal Title :Journal of the Korean Geotechnical Society
  • Journal Title(Ko) :한국지반공학회 논문집
  • Volume : 42
  • No :4
  • Pages :21-33
  • Received Date : 2026-05-29
  • Revised Date : 2026-06-29
  • Accepted Date : 2026-07-07
  • DOI :https://doi.org/10.7843/kgs.2026.42.4.21
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